diff --git "a/languages.json" "b/languages.json"
new file mode 100644--- /dev/null
+++ "b/languages.json"
@@ -0,0 +1,8138 @@
+[
+  {
+    "bcp_47":"en",
+    "speakers":1636485840,
+    "language_name":"English",
+    "autonym":"English",
+    "family":"Indo-European",
+    "flores_path":"eng_Latn",
+    "fleurs_tag":"en_us",
+    "commonvoice_hours":2661.0,
+    "commonvoice_locale":"en",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"zh",
+    "speakers":1304678914,
+    "language_name":"Chinese",
+    "autonym":"中文",
+    "family":"Sino-Tibetan",
+    "flores_path":"cmn_Hans",
+    "fleurs_tag":"cmn_hans_cn",
+    "commonvoice_hours":422.0,
+    "commonvoice_locale":"zh-TW",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"hi",
+    "speakers":546882144,
+    "language_name":"Hindi",
+    "autonym":"हिन्दी",
+    "family":"Indo-European",
+    "flores_path":"hin_Deva",
+    "fleurs_tag":"hi_in",
+    "commonvoice_hours":16.0,
+    "commonvoice_locale":"hi-IN",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"es",
+    "speakers":493528077,
+    "language_name":"Spanish",
+    "autonym":"Español",
+    "family":"Indo-European",
+    "flores_path":"spa_Latn",
+    "fleurs_tag":"es_419",
+    "commonvoice_hours":447.0,
+    "commonvoice_locale":"es",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ar",
+    "speakers":351664197,
+    "language_name":"Arabic",
+    "autonym":"العربية",
+    "family":"Afro-Asiatic",
+    "flores_path":"arb_Arab",
+    "fleurs_tag":"ar_eg",
+    "commonvoice_hours":92.0,
+    "commonvoice_locale":"ar",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ur",
+    "speakers":290790290,
+    "language_name":"Urdu",
+    "autonym":"اردو",
+    "family":"Indo-European",
+    "flores_path":"urd_Arab",
+    "fleurs_tag":"ur_pk",
+    "commonvoice_hours":77.0,
+    "commonvoice_locale":"ur",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"fr",
+    "speakers":278611507,
+    "language_name":"French",
+    "autonym":"Français",
+    "family":"Indo-European",
+    "flores_path":"fra_Latn",
+    "fleurs_tag":"fr_fr",
+    "commonvoice_hours":1057.0,
+    "commonvoice_locale":"fr",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"bn",
+    "speakers":267193288,
+    "language_name":"Bangla",
+    "autonym":"বাংলা",
+    "family":"Indo-European",
+    "flores_path":"ben_Beng",
+    "fleurs_tag":"bn_in",
+    "commonvoice_hours":49.0,
+    "commonvoice_locale":"bn",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"pt",
+    "speakers":237496885,
+    "language_name":"Portuguese",
+    "autonym":"Português",
+    "family":"Indo-European",
+    "flores_path":"por_Latn",
+    "fleurs_tag":"pt_br",
+    "commonvoice_hours":178.0,
+    "commonvoice_locale":"pt",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"pa",
+    "speakers":203571210,
+    "language_name":"Punjabi",
+    "autonym":"ਪੰਜਾਬੀ",
+    "family":"Indo-European",
+    "flores_path":"pan_Guru",
+    "fleurs_tag":"pa_in",
+    "commonvoice_hours":2.3,
+    "commonvoice_locale":"pa-IN",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ru",
+    "speakers":195841151,
+    "language_name":"Russian",
+    "autonym":"Русский",
+    "family":"Indo-European",
+    "flores_path":"rus_Cyrl",
+    "fleurs_tag":"ru_ru",
+    "commonvoice_hours":244.0,
+    "commonvoice_locale":"ru",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"sw",
+    "speakers":171610296,
+    "language_name":"Swahili",
+    "autonym":"Kiswahili",
+    "family":"Atlantic-Congo",
+    "flores_path":"swh_Latn",
+    "fleurs_tag":"sw_ke",
+    "commonvoice_hours":411.0,
+    "commonvoice_locale":"sw",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"id",
+    "speakers":171207687,
+    "language_name":"Indonesian",
+    "autonym":"Bahasa Indonesia",
+    "family":"Austronesian",
+    "flores_path":"ind_Latn",
+    "fleurs_tag":"id_id",
+    "commonvoice_hours":33.0,
+    "commonvoice_locale":"id",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"de",
+    "speakers":136350226,
+    "language_name":"German",
+    "autonym":"Deutsch",
+    "family":"Indo-European",
+    "flores_path":"deu_Latn",
+    "fleurs_tag":"de_de",
+    "commonvoice_hours":1362.0,
+    "commonvoice_locale":"de",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ja",
+    "speakers":119729026,
+    "language_name":"Japanese",
+    "autonym":"日本語",
+    "family":"Japonic",
+    "flores_path":"jpn_Jpan",
+    "fleurs_tag":"ja_jp",
+    "commonvoice_hours":223.0,
+    "commonvoice_locale":"ja",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"te",
+    "speakers":95478480,
+    "language_name":"Telugu",
+    "autonym":"తెలుగు",
+    "family":"Dravidian",
+    "flores_path":"tel_Telu",
+    "fleurs_tag":"te_in",
+    "commonvoice_hours":0.3,
+    "commonvoice_locale":"te",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"lah",
+    "speakers":93433552,
+    "language_name":"Western Panjabi",
+    "autonym":"لہندا پنجابی",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mr",
+    "speakers":92826300,
+    "language_name":"Marathi",
+    "autonym":"मराठी",
+    "family":"Indo-European",
+    "flores_path":"mar_Deva",
+    "fleurs_tag":"mr_in",
+    "commonvoice_hours":20.0,
+    "commonvoice_locale":"mr",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"jv",
+    "speakers":91180665,
+    "language_name":"Javanese",
+    "autonym":"Jawa",
+    "family":"Austronesian",
+    "flores_path":"jav_Latn",
+    "fleurs_tag":"jv_id",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"jv",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"vi",
+    "speakers":86222962,
+    "language_name":"Vietnamese",
+    "autonym":"Tiếng Việt",
+    "family":"Austroasiatic",
+    "flores_path":"vie_Latn",
+    "fleurs_tag":"vi_vn",
+    "commonvoice_hours":6.2,
+    "commonvoice_locale":"vi",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ta",
+    "speakers":85616159,
+    "language_name":"Tamil",
+    "autonym":"தமிழ்",
+    "family":"Dravidian",
+    "flores_path":"tam_Taml",
+    "fleurs_tag":"ta_in",
+    "commonvoice_hours":235.0,
+    "commonvoice_locale":"ta",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"fa",
+    "speakers":84710459,
+    "language_name":"Persian",
+    "autonym":"فارسی",
+    "family":"Indo-European",
+    "flores_path":"pes_Arab",
+    "fleurs_tag":"fa_ir",
+    "commonvoice_hours":370.0,
+    "commonvoice_locale":"fa",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"wuu",
+    "speakers":83641200,
+    "language_name":"Wu Chinese",
+    "autonym":"Wu Chinese",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tr",
+    "speakers":80360704,
+    "language_name":"Turkish",
+    "autonym":"Türkçe",
+    "family":"Turkic",
+    "flores_path":"tur_Latn",
+    "fleurs_tag":"tr_tr",
+    "commonvoice_hours":129.0,
+    "commonvoice_locale":"tr",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"yue",
+    "speakers":79654759,
+    "language_name":"Cantonese",
+    "autonym":"粵語",
+    "family":"Sino-Tibetan",
+    "flores_path":"yue_Hant",
+    "fleurs_tag":"yue_hant_hk",
+    "commonvoice_hours":203.0,
+    "commonvoice_locale":"yue",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ko",
+    "speakers":78357046,
+    "language_name":"Korean",
+    "autonym":"한국어",
+    "family":"Koreanic",
+    "flores_path":"kor_Hang",
+    "fleurs_tag":"ko_kr",
+    "commonvoice_hours":1.7,
+    "commonvoice_locale":"ko",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"it",
+    "speakers":70247060,
+    "language_name":"Italian",
+    "autonym":"Italiano",
+    "family":"Indo-European",
+    "flores_path":"ita_Latn",
+    "fleurs_tag":"it_it",
+    "commonvoice_hours":362.0,
+    "commonvoice_locale":"it",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"fil",
+    "speakers":67471096,
+    "language_name":"Filipino",
+    "autonym":"Filipino",
+    "family":"Austronesian",
+    "flores_path":"fil_Latn",
+    "fleurs_tag":"fil_ph",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"tl",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"arz",
+    "speakers":66639360,
+    "language_name":"Egyptian Arabic",
+    "autonym":"Egyptian Arabic",
+    "family":"Afro-Asiatic",
+    "flores_path":"arz_Arab",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"gu",
+    "speakers":61721799,
+    "language_name":"Gujarati",
+    "autonym":"ગુજરાતી",
+    "family":"Indo-European",
+    "flores_path":"guj_Gujr",
+    "fleurs_tag":"gu_in",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"gu-IN",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"th",
+    "speakers":55181920,
+    "language_name":"Thai",
+    "autonym":"ไทย",
+    "family":"Tai-Kadai",
+    "flores_path":"tha_Thai",
+    "fleurs_tag":"th_th",
+    "commonvoice_hours":172.0,
+    "commonvoice_locale":"th",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ps",
+    "speakers":53542641,
+    "language_name":"Pashto",
+    "autonym":"پښتو",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":"ps_af",
+    "commonvoice_hours":81.0,
+    "commonvoice_locale":"ps",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kn",
+    "speakers":49065330,
+    "language_name":"Kannada",
+    "autonym":"ಕನ್ನಡ",
+    "family":"Dravidian",
+    "flores_path":"kan_Knda",
+    "fleurs_tag":"kn_in",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"kn",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"pcm",
+    "speakers":44945880,
+    "language_name":"Nigerian Pidgin",
+    "autonym":"Naijíriá Píjin",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":14.0,
+    "commonvoice_locale":"pcm",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ml",
+    "speakers":43257484,
+    "language_name":"Malayalam",
+    "autonym":"മലയാ���ം",
+    "family":"Dravidian",
+    "flores_path":"mal_Mlym",
+    "fleurs_tag":"ml_in",
+    "commonvoice_hours":2.8,
+    "commonvoice_locale":"ml",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"or",
+    "speakers":42434880,
+    "language_name":"Odia",
+    "autonym":"ଓଡ଼ିଆ",
+    "family":"Indo-European",
+    "flores_path":"ory_Orya",
+    "fleurs_tag":"or_in",
+    "commonvoice_hours":2.8,
+    "commonvoice_locale":"or",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"pl",
+    "speakers":41077399,
+    "language_name":"Polish",
+    "autonym":"Polski",
+    "family":"Indo-European",
+    "flores_path":"pol_Latn",
+    "fleurs_tag":"pl_pl",
+    "commonvoice_hours":174.0,
+    "commonvoice_locale":"pl",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"hsn",
+    "speakers":40426580,
+    "language_name":"Xiang Chinese",
+    "autonym":"Xiang Chinese",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ha",
+    "speakers":40411882,
+    "language_name":"Hausa",
+    "autonym":"Hausa",
+    "family":"Afro-Asiatic",
+    "flores_path":"hau_Latn",
+    "fleurs_tag":"ha_ng",
+    "commonvoice_hours":4.1,
+    "commonvoice_locale":"ha",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"sd",
+    "speakers":40329510,
+    "language_name":"Sindhi",
+    "autonym":"سنڌي",
+    "family":"Indo-European",
+    "flores_path":"snd_Arab",
+    "fleurs_tag":"sd_in",
+    "commonvoice_hours":0.4,
+    "commonvoice_locale":"sd",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"apc",
+    "speakers":39031474,
+    "language_name":"North Levantine Arabic",
+    "autonym":"العامية",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ms",
+    "speakers":38097307,
+    "language_name":"Malay",
+    "autonym":"Bahasa Malaysia",
+    "family":"Austronesian",
+    "flores_path":"zsm_Latn",
+    "fleurs_tag":"ms_my",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ms",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"my",
+    "speakers":36559231,
+    "language_name":"Burmese",
+    "autonym":"မြန်မာ",
+    "family":"Sino-Tibetan",
+    "flores_path":"mya_Mymr",
+    "fleurs_tag":"my_mm",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"my",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"am",
+    "speakers":35728475,
+    "language_name":"Amharic",
+    "autonym":"አማርኛ",
+    "family":"Afro-Asiatic",
+    "flores_path":"amh_Ethi",
+    "fleurs_tag":"am_et",
+    "commonvoice_hours":1.8,
+    "commonvoice_locale":"am",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"arq",
+    "speakers":35667507,
+    "language_name":"Algerian Arabic",
+    "autonym":"Algerian Arabic",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"om",
+    "speakers":34897121,
+    "language_name":"Oromo",
+    "autonym":"Oromoo",
+    "family":"Afro-Asiatic",
+    "flores_path":"gaz_Latn",
+    "fleurs_tag":"om_et",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"om",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"bho",
+    "speakers":32934797,
+    "language_name":"Bhojpuri",
+    "autonym":"भोजपुरी",
+    "family":"Indo-European",
+    "flores_path":"bho_Deva",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"uz",
+    "speakers":32792780,
+    "language_name":"Uzbek",
+    "autonym":"O‘Zbek",
+    "family":"Turkic",
+    "flores_path":"uzn_Latn",
+    "fleurs_tag":"uz_uz",
+    "commonvoice_hours":100.0,
+    "commonvoice_locale":"uz",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"az",
+    "speakers":32446682,
+    "language_name":"Azerbaijani",
+    "autonym":"Azərbaycan",
+    "family":"Turkic",
+    "flores_path":"azj_Latn",
+    "fleurs_tag":"az_az",
+    "commonvoice_hours":0.5,
+    "commonvoice_locale":"az",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"hak",
+    "speakers":32062460,
+    "language_name":"Hakka Chinese",
+    "autonym":"Hakka Chinese",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"su",
+    "speakers":32043120,
+    "language_name":"Sundanese",
+    "autonym":"Basa Sunda",
+    "family":"Austronesian",
+    "flores_path":"sun_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"nl",
+    "speakers":31765645,
+    "language_name":"Dutch",
+    "autonym":"Nederlands",
+    "family":"Indo-European",
+    "flores_path":"nld_Latn",
+    "fleurs_tag":"nl_nl",
+    "commonvoice_hours":119.0,
+    "commonvoice_locale":"nl",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ary",
+    "speakers":30938679,
+    "language_name":"Moroccan Arabic",
+    "autonym":"Moroccan Arabic",
+    "family":"Afro-Asiatic",
+    "flores_path":"ary_Arab",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"uk",
+    "speakers":29348975,
+    "language_name":"Ukrainian",
+    "autonym":"Українська",
+    "family":"Indo-European",
+    "flores_path":"ukr_Cyrl",
+    "fleurs_tag":"uk_ua",
+    "commonvoice_hours":99.0,
+    "commonvoice_locale":"uk",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"yo",
+    "speakers":28685568,
+    "language_name":"Yoruba",
+    "autonym":"Èdè Yorùbá",
+    "family":"Atlantic-Congo",
+    "flores_path":"yor_Latn",
+    "fleurs_tag":"yo_ng",
+    "commonvoice_hours":6.1,
+    "commonvoice_locale":"yo",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"skr",
+    "speakers":28020120,
+    "language_name":"Saraiki",
+    "autonym":"سرائیکی",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":4.3,
+    "commonvoice_locale":"skr",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ig",
+    "speakers":27823640,
+    "language_name":"Igbo",
+    "autonym":"Igbo",
+    "family":"Atlantic-Congo",
+    "flores_path":"ibo_Latn",
+    "fleurs_tag":"ig_ng",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ig",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"nan",
+    "speakers":26486380,
+    "language_name":"Min Nan Chinese",
+    "autonym":"Min Nan Chinese",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ceb",
+    "speakers":26203440,
+    "language_name":"Cebuano",
+    "autonym":"Cebuano",
+    "family":"Austronesian",
+    "flores_path":"ceb_Latn",
+    "fleurs_tag":"ceb_ph",
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"awa",
+    "speakers":25862924,
+    "language_name":"Awadhi",
+    "autonym":"Awadhi",
+    "family":"Indo-European",
+    "flores_path":"awa_Deva",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"mg",
+    "speakers":24260130,
+    "language_name":"Malagasy",
+    "autonym":"Malagasy",
+    "family":"Austronesian",
+    "flores_path":"plt_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"mg",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"gan",
+    "speakers":23698340,
+    "language_name":"Gan Chinese",
+    "autonym":"Gan Chinese",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ro",
+    "speakers":22187408,
+    "language_name":"Romanian",
+    "autonym":"Română",
+    "family":"Indo-European",
+    "flores_path":"ron_Latn",
+    "fleurs_tag":"ro_ro",
+    "commonvoice_hours":21.0,
+    "commonvoice_locale":"ro",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"bar",
+    "speakers":22043627,
+    "language_name":"Bavarian",
+    "autonym":"Bavarian",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ne",
+    "speakers":20903374,
+    "language_name":"Nepali",
+    "autonym":"नेपाली",
+    "family":"Indo-European",
+    "flores_path":"npi_Deva",
+    "fleurs_tag":"ne_np",
+    "commonvoice_hours":1.3,
+    "commonvoice_locale":"ne-NP",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"mai",
+    "speakers":19249149,
+    "language_name":"Maithili",
+    "autonym":"मैथिली",
+    "family":"Indo-European",
+    "flores_path":"mai_Deva",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"mai",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"as",
+    "speakers":17239170,
+    "language_name":"Assamese",
+    "autonym":"অসমীয়া",
+    "family":"Indo-European",
+    "flores_path":"asm_Beng",
+    "fleurs_tag":"as_in",
+    "commonvoice_hours":2.8,
+    "commonvoice_locale":"as",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ny",
+    "speakers":17026781,
+    "language_name":"Nyanja",
+    "autonym":"Nyanja",
+    "family":"Atlantic-Congo",
+    "flores_path":"nya_Latn",
+    "fleurs_tag":"ny_mw",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ny",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"so",
+    "speakers":16911645,
+    "language_name":"Somali",
+    "autonym":"Soomaali",
+    "family":"Afro-Asiatic",
+    "flores_path":"som_Latn",
+    "fleurs_tag":"so_so",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"so",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"mad",
+    "speakers":16822638,
+    "language_name":"Madurese",
+    "autonym":"Madurese",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tts",
+    "speakers":16554576,
+    "language_name":"Northeastern Thai",
+    "autonym":"Northeastern Thai",
+    "family":"Tai-Kadai",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rkt",
+    "speakers":16274502,
+    "language_name":"Rangpuri",
+    "autonym":"Rangpuri",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mag",
+    "speakers":15913080,
+    "language_name":"Magahi",
+    "autonym":"Magahi",
+    "family":"Indo-European",
+    "flores_path":"mag_Deva",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"bgc",
+    "speakers":15913080,
+    "language_name":"Haryanvi",
+    "autonym":"हरियाणवी",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mwr",
+    "speakers":15913080,
+    "language_name":"Marwari",
+    "autonym":"Marwari",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sr",
+    "speakers":15602410,
+    "language_name":"Serbian",
+    "autonym":"Српски",
+    "family":"Indo-European",
+    "flores_path":"srp_Cyrl",
+    "fleurs_tag":"sr_rs",
+    "commonvoice_hours":7.5,
+    "commonvoice_locale":"sr",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"si",
+    "speakers":15564656,
+    "language_name":"Sinhala",
+    "autonym":"සිංහල",
+    "family":"Indo-European",
+    "flores_path":"sin_Sinh",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"si",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"km",
+    "speakers":15065030,
+    "language_name":"Khmer",
+    "autonym":"ខ្មែរ",
+    "family":"Austroasiatic",
+    "flores_path":"khm_Khmr",
+    "fleurs_tag":"km_kh",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"km",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"hne",
+    "speakers":14586990,
+    "language_name":"Chhattisgarhi",
+    "autonym":"Chhattisgarhi",
+    "family":"Indo-European",
+    "flores_path":"hne_Deva",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"fuv",
+    "speakers":14339876,
+    "language_name":"Nigerian Fulfulde",
+    "autonym":"Nigerian Fulfulde",
+    "family":"Atlantic-Congo",
+    "flores_path":"fuv_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"zu",
+    "speakers":13973830,
+    "language_name":"Zulu",
+    "autonym":"Isizulu",
+    "family":"Atlantic-Congo",
+    "flores_path":"zul_Latn",
+    "fleurs_tag":"zu_za",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"zu",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"kk",
+    "speakers":13637392,
+    "language_name":"Kazakh",
+    "autonym":"Қазақ Тілі",
+    "family":"Turkic",
+    "flores_path":"kaz_Cyrl",
+    "fleurs_tag":"kk_kz",
+    "commonvoice_hours":2.1,
+    "commonvoice_locale":"kk",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"dcc",
+    "speakers":13128291,
+    "language_name":"Deccan",
+    "autonym":"Deccan",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"cs",
+    "speakers":13045532,
+    "language_name":"Czech",
+    "autonym":"Čeština",
+    "family":"Indo-European",
+    "flores_path":"ces_Latn",
+    "fleurs_tag":"cs_cz",
+    "commonvoice_hours":75.0,
+    "commonvoice_locale":"cs",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"sv",
+    "speakers":12932871,
+    "language_name":"Swedish",
+    "autonym":"Svenska",
+    "family":"Indo-European",
+    "flores_path":"swe_Latn",
+    "fleurs_tag":"sv_se",
+    "commonvoice_hours":47.0,
+    "commonvoice_locale":"sv-SE",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"hu",
+    "speakers":12443430,
+    "language_name":"Hungarian",
+    "autonym":"Magyar",
+    "family":"Uralic",
+    "flores_path":"hun_Latn",
+    "fleurs_tag":"hu_hu",
+    "commonvoice_hours":92.0,
+    "commonvoice_locale":"hu",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"el",
+    "speakers":12292242,
+    "language_name":"Greek",
+    "autonym":"Ελληνικά",
+    "family":"Indo-European",
+    "flores_path":"ell_Grek",
+    "fleurs_tag":"el_gr",
+    "commonvoice_hours":20.0,
+    "commonvoice_locale":"el",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"sn",
+    "speakers":11782503,
+    "language_name":"Shona",
+    "autonym":"Chishona",
+    "family":"Atlantic-Congo",
+    "flores_path":"sna_Latn",
+    "fleurs_tag":"sn_zw",
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"nds",
+    "speakers":11520008,
+    "language_name":"Low German",
+    "autonym":"Neddersass’Sch",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ak",
+    "speakers":11442678,
+    "language_name":"Akan",
+    "autonym":"Akan",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.2,
+    "commonvoice_locale":"tw",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"qu",
+    "speakers":11385851,
+    "language_name":"Quechua",
+    "autonym":"Runasimi",
+    "family":"Quechuan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ckb",
+    "speakers":11086549,
+    "language_name":"Central Kurdish",
+    "autonym":"کوردیی ناوەندی",
+    "family":"Indo-European",
+    "flores_path":"ckb_Arab",
+    "fleurs_tag":"ckb_iq",
+    "commonvoice_hours":135.0,
+    "commonvoice_locale":"ckb",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"rw",
+    "speakers":11083625,
+    "language_name":"Kinyarwanda",
+    "autonym":"Kinyarwanda",
+    "family":"Atlantic-Congo",
+    "flores_path":"kin_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":2002.0,
+    "commonvoice_locale":"rw",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"wo",
+    "speakers":11025494,
+    "language_name":"Wolof",
+    "autonym":"Wolof",
+    "family":"Atlantic-Congo",
+    "flores_path":"wol_Latn",
+    "fleurs_tag":"wo_sn",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"wo",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"aeb",
+    "speakers":10549080,
+    "language_name":"Tunisian Arabic",
+    "autonym":"Tunisian Arabic",
+    "family":"Afro-Asiatic",
+    "flores_path":"aeb_Arab",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ilo",
+    "speakers":10481376,
+    "language_name":"Iloko",
+    "autonym":"Ilokano",
+    "family":"Austronesian",
+    "flores_path":"ilo_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"xh",
+    "speakers":10182944,
+    "language_name":"Xhosa",
+    "autonym":"Isixhosa",
+    "family":"Atlantic-Congo",
+    "flores_path":"xho_Latn",
+    "fleurs_tag":"xh_za",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"xh",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ti",
+    "speakers":10145911,
+    "language_name":"Tigrinya",
+    "autonym":"ትግርኛ",
+    "family":"Afro-Asiatic",
+    "flores_path":"tir_Ethi",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ti",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"be",
+    "speakers":10064517,
+    "language_name":"Belarusian",
+    "autonym":"Беларуская",
+    "family":"Indo-European",
+    "flores_path":"bel_Cyrl",
+    "fleurs_tag":"be_by",
+    "commonvoice_hours":1807.0,
+    "commonvoice_locale":"be",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"lua",
+    "speakers":9770880,
+    "language_name":"Luba-Lulua",
+    "autonym":"Luba-Lulua",
+    "family":"Atlantic-Congo",
+    "flores_path":"lua_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.1,
+    "commonvoice_locale":"lua",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"tg",
+    "speakers":9644223,
+    "language_name":"Tajik",
+    "autonym":"Тоҷикӣ",
+    "family":"Indo-European",
+    "flores_path":"tgk_Cyrl",
+    "fleurs_tag":"tg_tj",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"tg",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"umb",
+    "speakers":9431467,
+    "language_name":"Umbundu",
+    "autonym":"Umbundu",
+    "family":"Atlantic-Congo",
+    "flores_path":"umb_Latn",
+    "fleurs_tag":"umb_ao",
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"bm",
+    "speakers":9385632,
+    "language_name":"Bambara",
+    "autonym":"Bamanakan",
+    "family":"Mande",
+    "flores_path":"bam_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"bm",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"af",
+    "speakers":9318845,
+    "language_name":"Afrikaans",
+    "autonym":"Afrikaans",
+    "family":"Indo-European",
+    "flores_path":"afr_Latn",
+    "fleurs_tag":"af_za",
+    "commonvoice_hours":0.5,
+    "commonvoice_locale":"af",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"hil",
+    "speakers":9171204,
+    "language_name":"Hiligaynon",
+    "autonym":"Ilonggo",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"hil",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ki",
+    "speakers":9099743,
+    "language_name":"Kikuyu",
+    "autonym":"Gikuyu",
+    "family":"Atlantic-Congo",
+    "flores_path":"kik_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ki",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ht",
+    "speakers":8964918,
+    "language_name":"Haitian Creole",
+    "autonym":"Haitian Creole",
+    "family":"Indo-European",
+    "flores_path":"hat_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ht",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ca",
+    "speakers":8679139,
+    "language_name":"Catalan",
+    "autonym":"Català",
+    "family":"Indo-European",
+    "flores_path":"cat_Latn",
+    "fleurs_tag":"ca_es",
+    "commonvoice_hours":2857.0,
+    "commonvoice_locale":"ca",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"he",
+    "speakers":8675480,
+    "language_name":"Hebrew",
+    "autonym":"עברית",
+    "family":"Afro-Asiatic",
+    "flores_path":"heb_Hebr",
+    "fleurs_tag":"he_il",
+    "commonvoice_hours":1.2,
+    "commonvoice_locale":"he",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ii",
+    "speakers":8364120,
+    "language_name":"Sichuan Yi",
+    "autonym":"ꆈꌠꉙ",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mos",
+    "speakers":8334160,
+    "language_name":"Mossi",
+    "autonym":"Mossi",
+    "family":"Atlantic-Congo",
+    "flores_path":"mos_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"mos",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"bal",
+    "speakers":8227887,
+    "language_name":"Baluchi",
+    "autonym":"بلۆچی",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"bal",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"syl",
+    "speakers":8132550,
+    "language_name":"Sylheti",
+    "autonym":"Sylheti",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kmb",
+    "speakers":8130575,
+    "language_name":"Kimbundu",
+    "autonym":"Kimbundu",
+    "family":"Atlantic-Congo",
+    "flores_path":"kmb_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ug",
+    "speakers":8052967,
+    "language_name":"Uyghur",
+    "autonym":"ئۇيغۇرچە",
+    "family":"Turkic",
+    "flores_path":"uig_Arab",
+    "fleurs_tag":null,
+    "commonvoice_hours":370.0,
+    "commonvoice_locale":"ug",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"min",
+    "speakers":8010780,
+    "language_name":"Minangkabau",
+    "autonym":"Minangkabau",
+    "family":"Austronesian",
+    "flores_path":"min_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"gsw",
+    "speakers":7956952,
+    "language_name":"Swiss German",
+    "autonym":"Schwiizertüütsch",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"gsw",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bg",
+    "speakers":7878315,
+    "language_name":"Bulgarian",
+    "autonym":"Български",
+    "family":"Indo-European",
+    "flores_path":"bul_Cyrl",
+    "fleurs_tag":"bg_bg",
+    "commonvoice_hours":16.0,
+    "commonvoice_locale":"bg",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"zgh",
+    "speakers":7823574,
+    "language_name":"Standard Moroccan Tamazight",
+    "autonym":"ⵜⴰⵎⴰⵣⵉⵖⵜ",
+    "family":"Afro-Asiatic",
+    "flores_path":"zgh_Tfng",
+    "fleurs_tag":null,
+    "commonvoice_hours":1.3,
+    "commonvoice_locale":"zgh",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ff",
+    "speakers":7788904,
+    "language_name":"Fula",
+    "autonym":"Pulaar",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":"ff_sn",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ff",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bs",
+    "speakers":7594468,
+    "language_name":"Bosnian",
+    "autonym":"Bosanski",
+    "family":"Indo-European",
+    "flores_path":"bos_Latn",
+    "fleurs_tag":"bs_ba",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"bs",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"rn",
+    "speakers":7475454,
+    "language_name":"Rundi",
+    "autonym":"Ikirundi",
+    "family":"Atlantic-Congo",
+    "flores_path":"run_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"rn",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"bjj",
+    "speakers":7426104,
+    "language_name":"Kanauji",
+    "autonym":"Kanauji",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sat",
+    "speakers":7293495,
+    "language_name":"Santali",
+    "autonym":"ᱥᱟᱱᱛᱟᱲᱤ",
+    "family":"Austroasiatic",
+    "flores_path":"sat_Olck",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.5,
+    "commonvoice_locale":"sat",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"da",
+    "speakers":7072056,
+    "language_name":"Danish",
+    "autonym":"Dansk",
+    "family":"Indo-European",
+    "flores_path":"dan_Latn",
+    "fleurs_tag":"da_dk",
+    "commonvoice_hours":13.0,
+    "commonvoice_locale":"da",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"tk",
+    "speakers":6870838,
+    "language_name":"Turkmen",
+    "autonym":"Türkmen Dili",
+    "family":"Turkic",
+    "flores_path":"tuk_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":2.8,
+    "commonvoice_locale":"tk",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ku",
+    "speakers":6866757,
+    "language_name":"Kurdish",
+    "autonym":"Kurdî (Kurmancî)",
+    "family":"Indo-European",
+    "flores_path":"kmr_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":69.0,
+    "commonvoice_locale":"kmr",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"hr",
+    "speakers":6813164,
+    "language_name":"Croatian",
+    "autonym":"Hrvatski",
+    "family":"Indo-European",
+    "flores_path":"hrv_Latn",
+    "fleurs_tag":"hr_hr",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"hr",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"sq",
+    "speakers":6791906,
+    "language_name":"Albanian",
+    "autonym":"Shqip",
+    "family":"Indo-European",
+    "flores_path":"als_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":8.8,
+    "commonvoice_locale":"sq",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"sk",
+    "speakers":6680269,
+    "language_name":"Slovak",
+    "autonym":"Slovenčina",
+    "family":"Indo-European",
+    "flores_path":"slk_Latn",
+    "fleurs_tag":"sk_sk",
+    "commonvoice_hours":48.0,
+    "commonvoice_locale":"sk",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"dyu",
+    "speakers":6667328,
+    "language_name":"Dyula",
+    "autonym":"Dyula",
+    "family":"Mande",
+    "flores_path":"dyu_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.4,
+    "commonvoice_locale":"dyu",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"nod",
+    "speakers":6621830,
+    "language_name":"Northern Thai",
+    "autonym":"Northern Thai",
+    "family":"Tai-Kadai",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mn",
+    "speakers":6572846,
+    "language_name":"Mongolian",
+    "autonym":"Монгол",
+    "family":"Mongolic-Khitan",
+    "flores_path":"khk_Cyrl",
+    "fleurs_tag":"mn_mn",
+    "commonvoice_hours":46.0,
+    "commonvoice_locale":"mn",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"st",
+    "speakers":6390567,
+    "language_name":"Southern Sotho",
+    "autonym":"Sesotho",
+    "family":"Atlantic-Congo",
+    "flores_path":"sot_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"st",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"kri",
+    "speakers":6293684,
+    "language_name":"Krio",
+    "autonym":"Krio",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"shi",
+    "speakers":6187736,
+    "language_name":"Tachelhit",
+    "autonym":"ⵜⴰⵛⵍⵃⵉⵜ",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"shi",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tn",
+    "speakers":6113428,
+    "language_name":"Tswana",
+    "autonym":"Tswana",
+    "family":"Atlantic-Congo",
+    "flores_path":"tsn_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":4.2,
+    "commonvoice_locale":"tn",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"wtm",
+    "speakers":6100014,
+    "language_name":"Mewati",
+    "autonym":"Mewati",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"luy",
+    "speakers":5888069,
+    "language_name":"Luyia",
+    "autonym":"Luluhia",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"bxk",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gn",
+    "speakers":5827107,
+    "language_name":"Guarani",
+    "autonym":"Avañe’Ẽ",
+    "family":"Tupian",
+    "flores_path":"gug_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":3.7,
+    "commonvoice_locale":"gn",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"fi",
+    "speakers":5736842,
+    "language_name":"Finnish",
+    "autonym":"Suomi",
+    "family":"Uralic",
+    "flores_path":"fin_Latn",
+    "fleurs_tag":"fi_fi",
+    "commonvoice_hours":15.0,
+    "commonvoice_locale":"fi",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"lg",
+    "speakers":5622890,
+    "language_name":"Ganda",
+    "autonym":"Luganda",
+    "family":"Atlantic-Congo",
+    "flores_path":"lug_Latn",
+    "fleurs_tag":"lg_ug",
+    "commonvoice_hours":437.0,
+    "commonvoice_locale":"lg",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"bew",
+    "speakers":5607546,
+    "language_name":"Betawi",
+    "autonym":"Betawi",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"bew",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ks",
+    "speakers":5598085,
+    "language_name":"Kashmiri",
+    "autonym":"کٲشُر",
+    "family":"Indo-European",
+    "flores_path":"kas_Arab",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ks",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"sou",
+    "speakers":5518192,
+    "language_name":"Southern Thai",
+    "autonym":"Southern Thai",
+    "family":"Tai-Kadai",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nb",
+    "speakers":5468932,
+    "language_name":"Norwegian Bokmål",
+    "autonym":"Norsk Bokmål",
+    "family":"Indo-European",
+    "flores_path":"nob_Latn",
+    "fleurs_tag":"nb_no",
+    "commonvoice_hours":0.3,
+    "commonvoice_locale":"nb-NO",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"no",
+    "speakers":5467440,
+    "language_name":"Norwegian",
+    "autonym":"Norsk",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bem",
+    "speakers":5402246,
+    "language_name":"Bemba",
+    "autonym":"Ichibemba",
+    "family":"Atlantic-Congo",
+    "flores_path":"bem_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"hy",
+    "speakers":5317273,
+    "language_name":"Armenian",
+    "autonym":"Հայերեն",
+    "family":"Indo-European",
+    "flores_path":"hye_Armn",
+    "fleurs_tag":"hy_am",
+    "commonvoice_hours":31.0,
+    "commonvoice_locale":"hy-AM",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"nso",
+    "speakers":5307578,
+    "language_name":"Northern Sotho",
+    "autonym":"Northern Sotho",
+    "family":"Atlantic-Congo",
+    "flores_path":"nso_Latn",
+    "fleurs_tag":"nso_za",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"nso",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"luo",
+    "speakers":5245734,
+    "language_name":"Luo (Kenya and Tanzania)",
+    "autonym":"Dholuo",
+    "family":"Nilotic",
+    "flores_path":"luo_Latn",
+    "fleurs_tag":"luo_ke",
+    "commonvoice_hours":30.0,
+    "commonvoice_locale":"luo",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"tpi",
+    "speakers":5154217,
+    "language_name":"Tok Pisin",
+    "autonym":"Tok Pisin",
+    "family":"Indo-European",
+    "flores_path":"tpi_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"lo",
+    "speakers":5138706,
+    "language_name":"Lao",
+    "autonym":"ລາວ",
+    "family":"Tai-Kadai",
+    "flores_path":"lao_Laoo",
+    "fleurs_tag":"lo_la",
+    "commonvoice_hours":0.3,
+    "commonvoice_locale":"lo",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"suk",
+    "speakers":5094094,
+    "language_name":"Sukuma",
+    "autonym":"Sukuma",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kok",
+    "speakers":4906533,
+    "language_name":"Konkani",
+    "autonym":"कोंकणी",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"knn",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ts",
+    "speakers":4880932,
+    "language_name":"Tsonga",
+    "autonym":"Tsonga",
+    "family":"Atlantic-Congo",
+    "flores_path":"tso_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ts",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"vmf",
+    "speakers":4809582,
+    "language_name":"Main-Franconian",
+    "autonym":"Main-Franconian",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ban",
+    "speakers":4806468,
+    "language_name":"Balinese",
+    "autonym":"Balinese",
+    "family":"Austronesian",
+    "flores_path":"ban_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ee",
+    "speakers":4690857,
+    "language_name":"Ewe",
+    "autonym":"Eʋegbe",
+    "family":"Atlantic-Congo",
+    "flores_path":"ewe_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ee",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"za",
+    "speakers":4321462,
+    "language_name":"Zhuang",
+    "autonym":"Vahcuengh",
+    "family":"Tai-Kadai",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bug",
+    "speakers":4298211,
+    "language_name":"Buginese",
+    "autonym":"Buginese",
+    "family":"Austronesian",
+    "flores_path":"bug_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"mzn",
+    "speakers":4246165,
+    "language_name":"Mazanderani",
+    "autonym":"مازرونی",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gom",
+    "speakers":4243488,
+    "language_name":"Goan Konkani",
+    "autonym":"Goan Konkani",
+    "family":"Indo-European",
+    "flores_path":"gom_Deva",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"gom",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"kam",
+    "speakers":4068120,
+    "language_name":"Kamba",
+    "autonym":"Kikamba",
+    "family":"Atlantic-Congo",
+    "flores_path":"kam_Latn",
+    "fleurs_tag":"kam_ke",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"kam",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"kln",
+    "speakers":4068120,
+    "language_name":"Kalenjin",
+    "autonym":"Kalenjin",
+    "family":"Nilotic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":43.0,
+    "commonvoice_locale":"kln",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bjn",
+    "speakers":4010288,
+    "language_name":"Banjar",
+    "autonym":"Banjar",
+    "family":"Austronesian",
+    "flores_path":"bjn_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"hno",
+    "speakers":3969517,
+    "language_name":"Northern Hindko",
+    "autonym":"Northern Hindko",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":11.0,
+    "commonvoice_locale":"hno",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"vmw",
+    "speakers":3912766,
+    "language_name":"Makhuwa",
+    "autonym":"Emakhuwa",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"vmw",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"glk",
+    "speakers":3906472,
+    "language_name":"Gilaki",
+    "autonym":"Gilaki",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lmo",
+    "speakers":3901518,
+    "language_name":"Lombard",
+    "autonym":"Lombard",
+    "family":"Indo-European",
+    "flores_path":"lmo_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"dje",
+    "speakers":3871308,
+    "language_name":"Zarma",
+    "autonym":"Zarmaciine",
+    "family":"Songhay",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ndc",
+    "speakers":3867046,
+    "language_name":"Ndau",
+    "autonym":"Ndau",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sid",
+    "speakers":3783955,
+    "language_name":"Sidamo",
+    "autonym":"Sidamo",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ace",
+    "speakers":3738364,
+    "language_name":"Achinese",
+    "autonym":"Achinese",
+    "family":"Austronesian",
+    "flores_path":"ace_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ace",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"swv",
+    "speakers":3713052,
+    "language_name":"Shekhawati",
+    "autonym":"Shekhawati",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rif",
+    "speakers":3692411,
+    "language_name":"Riffian",
+    "autonym":"Tarifit",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"rif",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"shn",
+    "speakers":3687984,
+    "language_name":"Shan",
+    "autonym":"တႆး",
+    "family":"Tai-Kadai",
+    "flores_path":"shn_Mymr",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"shn",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"lmn",
+    "speakers":3580443,
+    "language_name":"Lambadi",
+    "autonym":"Lambadi",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gbm",
+    "speakers":3580443,
+    "language_name":"Garhwali",
+    "autonym":"Garhwali",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ka",
+    "speakers":3543646,
+    "language_name":"Georgian",
+    "autonym":"ქართული",
+    "family":"Kartvelian",
+    "flores_path":"kat_Geor",
+    "fleurs_tag":"ka_ge",
+    "commonvoice_hours":164.0,
+    "commonvoice_locale":"ka",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"gl",
+    "speakers":3515530,
+    "language_name":"Galician",
+    "autonym":"Galego",
+    "family":"Indo-European",
+    "flores_path":"glg_Latn",
+    "fleurs_tag":"gl_es",
+    "commonvoice_hours":115.0,
+    "commonvoice_locale":"gl",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ln",
+    "speakers":3514491,
+    "language_name":"Lingala",
+    "autonym":"Lingála",
+    "family":"Atlantic-Congo",
+    "flores_path":"lin_Latn",
+    "fleurs_tag":"ln_cd",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ln",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"man",
+    "speakers":3511762,
+    "language_name":"Mandingo",
+    "autonym":"Mandingo",
+    "family":"Mande",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tzm",
+    "speakers":3485047,
+    "language_name":"Central Atlas Tamazight",
+    "autonym":"Tamaziɣt N Laṭlaṣ",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mfa",
+    "speakers":3448870,
+    "language_name":"Pattani Malay",
+    "autonym":"Pattani Malay",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tiv",
+    "speakers":3424448,
+    "language_name":"Tiv",
+    "autonym":"Tiv",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kab",
+    "speakers":3351886,
+    "language_name":"Kabyle",
+    "autonym":"Taqbaylit",
+    "family":"Afro-Asiatic",
+    "flores_path":"kab_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":571.0,
+    "commonvoice_locale":"kab",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ky",
+    "speakers":3338267,
+    "language_name":"Kyrgyz",
+    "autonym":"Кыргызча",
+    "family":"Turkic",
+    "flores_path":"kir_Cyrl",
+    "fleurs_tag":"ky_kg",
+    "commonvoice_hours":39.0,
+    "commonvoice_locale":"ky",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"bik",
+    "speakers":3275430,
+    "language_name":"Bikol",
+    "autonym":"Bikol",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"fon",
+    "speakers":3216150,
+    "language_name":"Fon",
+    "autonym":"Fon",
+    "family":"Atlantic-Congo",
+    "flores_path":"fon_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"gon",
+    "speakers":3182616,
+    "language_name":"Gondi",
+    "autonym":"Gondi",
+    "family":"Dravidian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"war",
+    "speakers":3166927,
+    "language_name":"Waray",
+    "autonym":"Waray",
+    "family":"Austronesian",
+    "flores_path":"war_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"sdh",
+    "speakers":3142162,
+    "language_name":"Southern Kurdish",
+    "autonym":"کوردی خوارگ",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"sdh",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"brh",
+    "speakers":3035513,
+    "language_name":"Brahui",
+    "autonym":"Brahui",
+    "family":"Dravidian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bci",
+    "speakers":3022921,
+    "language_name":"Baoulé",
+    "autonym":"Baoulé",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":2.3,
+    "commonvoice_locale":"bci",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bo",
+    "speakers":3006697,
+    "language_name":"Tibetan",
+    "autonym":"བོད་སྐད་",
+    "family":"Sino-Tibetan",
+    "flores_path":"bod_Tibt",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"bo",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ibb",
+    "speakers":2996392,
+    "language_name":"Ibibio",
+    "autonym":"Ibibio",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":4.1,
+    "commonvoice_locale":"ibb",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"efi",
+    "speakers":2996392,
+    "language_name":"Efik",
+    "autonym":"Efik",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sg",
+    "speakers":2935521,
+    "language_name":"Sango",
+    "autonym":"Sängö",
+    "family":"Atlantic-Congo",
+    "flores_path":"sag_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"kfy",
+    "speakers":2917398,
+    "language_name":"Kumaoni",
+    "autonym":"Kumaoni",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ay",
+    "speakers":2838620,
+    "language_name":"Aymara",
+    "autonym":"Aymara",
+    "family":"Aymaran",
+    "flores_path":"ayr_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"nyn",
+    "speakers":2724939,
+    "language_name":"Nyankole",
+    "autonym":"Runyankore",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"nyn",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"jam",
+    "speakers":2668142,
+    "language_name":"Jamaican Creole English",
+    "autonym":"Jamaican Creole English",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"jam",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"doi",
+    "speakers":2652180,
+    "language_name":"Dogri",
+    "autonym":"डोगरी",
+    "family":"Indo-European",
+    "flores_path":"dgo_Deva",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"guz",
+    "speakers":2622867,
+    "language_name":"Gusii",
+    "autonym":"Ekegusii",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sas",
+    "speakers":2590152,
+    "language_name":"Sasak",
+    "autonym":"Sasak",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kru",
+    "speakers":2519571,
+    "language_name":"Kurukh",
+    "autonym":"Kurukh",
+    "family":"Dravidian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"pam",
+    "speakers":2511163,
+    "language_name":"Pampanga",
+    "autonym":"Pampanga",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"fbl",
+    "speakers":2511163,
+    "language_name":"West Albay Bikol",
+    "autonym":"West Albay Bikol",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lt",
+    "speakers":2488617,
+    "language_name":"Lithuanian",
+    "autonym":"Lietuvių",
+    "family":"Indo-European",
+    "flores_path":"lit_Latn",
+    "fleurs_tag":"lt_lt",
+    "commonvoice_hours":25.0,
+    "commonvoice_locale":"lt",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"bej",
+    "speakers":2460326,
+    "language_name":"Beja",
+    "autonym":"Beja",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bbc",
+    "speakers":2456639,
+    "language_name":"Batak Toba",
+    "autonym":"Batak Toba",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sck",
+    "speakers":2386962,
+    "language_name":"Sadri",
+    "autonym":"Sadri",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"wbq",
+    "speakers":2386962,
+    "language_name":"Waddar",
+    "autonym":"Waddar",
+    "family":"Dravidian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lu",
+    "speakers":2340940,
+    "language_name":"Luba-Katanga",
+    "autonym":"Tshiluba",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"cgg",
+    "speakers":2335662,
+    "language_name":"Chiga",
+    "autonym":"Rukiga",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"cgg",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"xog",
+    "speakers":2292409,
+    "language_name":"Soga",
+    "autonym":"Olusoga",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ss",
+    "speakers":2212379,
+    "language_name":"Swati",
+    "autonym":"Siswati",
+    "family":"Atlantic-Congo",
+    "flores_path":"ssw_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ss",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"haz",
+    "speakers":2161984,
+    "language_name":"Hazaragi",
+    "autonym":"Hazaragi",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mer",
+    "speakers":2141116,
+    "language_name":"Meru",
+    "autonym":"Kĩmĩrũ",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"xnr",
+    "speakers":2121744,
+    "language_name":"Kangri",
+    "autonym":"कांगड़ी",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"aa",
+    "speakers":2119663,
+    "language_name":"Afar",
+    "autonym":"Afar",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"aa",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"teo",
+    "speakers":2082973,
+    "language_name":"Teso",
+    "autonym":"Kiteso",
+    "family":"Nilotic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ngl",
+    "speakers":2046678,
+    "language_name":"Lomwe",
+    "autonym":"Lomwe",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"oc",
+    "speakers":2040398,
+    "language_name":"Occitan",
+    "autonym":"Occitan",
+    "family":"Indo-European",
+    "flores_path":"oci_Latn",
+    "fleurs_tag":"oc_fr",
+    "commonvoice_hours":1.8,
+    "commonvoice_locale":"oc",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"bgn",
+    "speakers":2037382,
+    "language_name":"Western Balochi",
+    "autonym":"بلوچی (رخشانی)",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lrc",
+    "speakers":2020512,
+    "language_name":"Northern Luri",
+    "autonym":"لۊری شومالی",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"wbr",
+    "speakers":1989135,
+    "language_name":"Wagdi",
+    "autonym":"Wagdi",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tcy",
+    "speakers":1989135,
+    "language_name":"Tulu",
+    "autonym":"Tulu",
+    "family":"Dravidian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"khn",
+    "speakers":1989135,
+    "language_name":"Khandesi",
+    "autonym":"Khandesi",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tt",
+    "speakers":1984108,
+    "language_name":"Tatar",
+    "autonym":"Татар",
+    "family":"Turkic",
+    "flores_path":"tat_Cyrl",
+    "fleurs_tag":null,
+    "commonvoice_hours":32.0,
+    "commonvoice_locale":"tt",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"sl",
+    "speakers":1973181,
+    "language_name":"Slovenian",
+    "autonym":"Slovenščina",
+    "family":"Indo-European",
+    "flores_path":"slv_Latn",
+    "fleurs_tag":"sl_si",
+    "commonvoice_hours":18.0,
+    "commonvoice_locale":"sl",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"mak",
+    "speakers":1949290,
+    "language_name":"Makasar",
+    "autonym":"Makasar",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"wal",
+    "speakers":1946034,
+    "language_name":"Wolaytta",
+    "autonym":"Wolaytta",
+    "family":"Ta-Ne-Omotic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nym",
+    "speakers":1932242,
+    "language_name":"Nyamwezi",
+    "autonym":"Nyamwezi",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"brx",
+    "speakers":1856526,
+    "language_name":"Bodo",
+    "autonym":"बर’",
+    "family":"Sino-Tibetan",
+    "flores_path":"brx_Deva",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ljp",
+    "speakers":1842479,
+    "language_name":"Lampung Api",
+    "autonym":"Lampung Api",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"cv",
+    "speakers":1842386,
+    "language_name":"Chuvash",
+    "autonym":"Чӑваш",
+    "family":"Turkic",
+    "flores_path":"chv_Cyrl",
+    "fleurs_tag":null,
+    "commonvoice_hours":27.0,
+    "commonvoice_locale":"cv",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ba",
+    "speakers":1842386,
+    "language_name":"Bashkir",
+    "autonym":"Башҡорт Теле",
+    "family":"Turkic",
+    "flores_path":"bak_Cyrl",
+    "fleurs_tag":null,
+    "commonvoice_hours":259.0,
+    "commonvoice_locale":"ba",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"rhg",
+    "speakers":1824082,
+    "language_name":"Rohingya",
+    "autonym":"𐴌𐴗𐴥𐴝𐴙𐴚𐴒𐴙𐴝",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"men",
+    "speakers":1813083,
+    "language_name":"Mende",
+    "autonym":"Mende",
+    "family":"Mande",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tum",
+    "speakers":1780514,
+    "language_name":"Tumbuka",
+    "autonym":"Tumbuka",
+    "family":"Atlantic-Congo",
+    "flores_path":"tum_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"tmh",
+    "speakers":1776965,
+    "language_name":"Tamashek",
+    "autonym":"Tamashek",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nd",
+    "speakers":1745556,
+    "language_name":"North Ndebele",
+    "autonym":"Isindebele",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"nd",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mas",
+    "speakers":1734738,
+    "language_name":"Masai",
+    "autonym":"Maa",
+    "family":"Nilotic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"srr",
+    "speakers":1731004,
+    "language_name":"Serer",
+    "autonym":"Serer",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"noe",
+    "speakers":1723917,
+    "language_name":"Nimadi",
+    "autonym":"Nimadi",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tem",
+    "speakers":1722482,
+    "language_name":"Timne",
+    "autonym":"Timne",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sco",
+    "speakers":1644028,
+    "language_name":"Scots",
+    "autonym":"Scots",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"sco",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"laj",
+    "speakers":1643614,
+    "language_name":"Lango (Uganda)",
+    "autonym":"Lango (Uganda)",
+    "family":"Nilotic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rmt",
+    "speakers":1613543,
+    "language_name":"Domari",
+    "autonym":"Domari",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mk",
+    "speakers":1608565,
+    "language_name":"Macedonian",
+    "autonym":"Македонски",
+    "family":"Indo-European",
+    "flores_path":"mkd_Cyrl",
+    "fleurs_tag":"mk_mk",
+    "commonvoice_hours":19.0,
+    "commonvoice_locale":"mk",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ach",
+    "speakers":1600361,
+    "language_name":"Acoli",
+    "autonym":"Acoli",
+    "family":"Nilotic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"fuq",
+    "speakers":1594068,
+    "language_name":"Central-Eastern Niger Fulfulde",
+    "autonym":"Central-Eastern Niger Fulfulde",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bhb",
+    "speakers":1591308,
+    "language_name":"Bhili",
+    "autonym":"Bhili",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"pag",
+    "speakers":1528534,
+    "language_name":"Pangasinan",
+    "autonym":"Pangasinan",
+    "family":"Austronesian",
+    "flores_path":"pag_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"kg",
+    "speakers":1526700,
+    "language_name":"Kongo",
+    "autonym":"Kongo",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bin",
+    "speakers":1519599,
+    "language_name":"Bini",
+    "autonym":"Bini",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ffm",
+    "speakers":1505612,
+    "language_name":"Maasina Fulfulde",
+    "autonym":"Maasina Fulfulde",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mni",
+    "speakers":1476591,
+    "language_name":"Manipuri",
+    "autonym":"মৈতৈলোন্",
+    "family":"Sino-Tibetan",
+    "flores_path":"mni_Beng",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"mni",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"abr",
+    "speakers":1467010,
+    "language_name":"Abron",
+    "autonym":"Abron",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kde",
+    "speakers":1463820,
+    "language_name":"Makonde",
+    "autonym":"Chimakonde",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"aln",
+    "speakers":1430250,
+    "language_name":"Gheg Albanian",
+    "autonym":"Gheg Albanian",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"aln",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ve",
+    "speakers":1391759,
+    "language_name":"Venda",
+    "autonym":"Tshivenḓa",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ve",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"seh",
+    "speakers":1384517,
+    "language_name":"Sena",
+    "autonym":"Sena",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"vec",
+    "speakers":1380829,
+    "language_name":"Venetian",
+    "autonym":"Veneto",
+    "family":"Indo-European",
+    "flores_path":"vec_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"vec",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"sus",
+    "speakers":1378014,
+    "language_name":"Susu",
+    "autonym":"Susu",
+    "family":"Mande",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nn",
+    "speakers":1366860,
+    "language_name":"Norwegian Nynorsk",
+    "autonym":"Norsk Nynorsk",
+    "family":"Indo-European",
+    "flores_path":"nno_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":1.5,
+    "commonvoice_locale":"nn-NO",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"mgh",
+    "speakers":1354419,
+    "language_name":"Makhuwa-Meetto",
+    "autonym":"Makua",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"raj",
+    "speakers":1326090,
+    "language_name":"Rajasthani",
+    "autonym":"राजस्थानी",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"hoc",
+    "speakers":1312829,
+    "language_name":"Ho",
+    "autonym":"Ho",
+    "family":"Austroasiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mdh",
+    "speakers":1310172,
+    "language_name":"Maguindanaon",
+    "autonym":"Maguindanaon",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mtr",
+    "speakers":1286307,
+    "language_name":"Mewari",
+    "autonym":"Mewari",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bum",
+    "speakers":1276270,
+    "language_name":"Bulu",
+    "autonym":"Bulu",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":9.8,
+    "commonvoice_locale":"bum",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"myx",
+    "speakers":1254337,
+    "language_name":"Masaaba",
+    "autonym":"Masaaba",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"unr",
+    "speakers":1252287,
+    "language_name":"Mundari",
+    "autonym":"Mundari",
+    "family":"Austroasiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mfe",
+    "speakers":1241433,
+    "language_name":"Morisyen",
+    "autonym":"Kreol Morisien",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ga",
+    "speakers":1237487,
+    "language_name":"Irish",
+    "autonym":"Gaeilge",
+    "family":"Indo-European",
+    "flores_path":"gle_Latn",
+    "fleurs_tag":"ga_ie",
+    "commonvoice_hours":6.2,
+    "commonvoice_locale":"ga-IE",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"fvr",
+    "speakers":1230163,
+    "language_name":"Fur",
+    "autonym":"Fur",
+    "family":"Furan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rej",
+    "speakers":1228320,
+    "language_name":"Rejang",
+    "autonym":"Rejang",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bhi",
+    "speakers":1220003,
+    "language_name":"Bhilali",
+    "autonym":"Bhilali",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tsg",
+    "speakers":1200991,
+    "language_name":"Tausug",
+    "autonym":"Tausug",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"quc",
+    "speakers":1200731,
+    "language_name":"Kʼicheʼ",
+    "autonym":"KʼIcheʼ",
+    "family":"Mayan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"quc",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bqi",
+    "speakers":1188926,
+    "language_name":"Bakhtiari",
+    "autonym":"Bakhtiari",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kpe",
+    "speakers":1186303,
+    "language_name":"Kpelle",
+    "autonym":"Kpɛlɛɛ",
+    "family":"Mande",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sef",
+    "speakers":1181687,
+    "language_name":"Cebaara Senoufo",
+    "autonym":"Cebaara Senoufo",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kxm",
+    "speakers":1172616,
+    "language_name":"Northern Khmer",
+    "autonym":"Northern Khmer",
+    "family":"Austroasiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"vls",
+    "speakers":1172070,
+    "language_name":"West Flemish",
+    "autonym":"West Flemish",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"snk",
+    "speakers":1153651,
+    "language_name":"Soninke",
+    "autonym":"Soninke",
+    "family":"Mande",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"snk",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"zza",
+    "speakers":1148245,
+    "language_name":"Zaza",
+    "autonym":"Zaza",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":1.8,
+    "commonvoice_locale":"zza",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lv",
+    "speakers":1147550,
+    "language_name":"Latvian",
+    "autonym":"Latviešu",
+    "family":"Indo-European",
+    "flores_path":"lvs_Latn",
+    "fleurs_tag":"lv_lv",
+    "commonvoice_hours":261.0,
+    "commonvoice_locale":"lv",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"dnj",
+    "speakers":1099244,
+    "language_name":"Dan",
+    "autonym":"Dan",
+    "family":"Mande",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gor",
+    "speakers":1094807,
+    "language_name":"Gorontalo",
+    "autonym":"Gorontalo",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tig",
+    "speakers":1094616,
+    "language_name":"Tigre",
+    "autonym":"Tigre",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":12.0,
+    "commonvoice_locale":"tig",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"eu",
+    "speakers":1088519,
+    "language_name":"Basque",
+    "autonym":"Euskara",
+    "family":null,
+    "flores_path":"eus_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":344.0,
+    "commonvoice_locale":"eu",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"hoj",
+    "speakers":1087394,
+    "language_name":"Hadothi",
+    "autonym":"Hadothi",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kbd",
+    "speakers":1070873,
+    "language_name":"Kabardian",
+    "autonym":"Kabardian",
+    "family":"Abkhaz-Adyge",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":51.0,
+    "commonvoice_locale":"kbd",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kha",
+    "speakers":1060872,
+    "language_name":"Khasi",
+    "autonym":"Khasi",
+    "family":"Austroasiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sc",
+    "speakers":1060846,
+    "language_name":"Sardinian",
+    "autonym":"Sardu",
+    "family":"Indo-European",
+    "flores_path":"srd_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":2.9,
+    "commonvoice_locale":"sc",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"loz",
+    "speakers":1045596,
+    "language_name":"Lozi",
+    "autonym":"Lozi",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gur",
+    "speakers":1026907,
+    "language_name":"Frafra",
+    "autonym":"Frafra",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ars",
+    "speakers":1025205,
+    "language_name":"Najdi Arabic",
+    "autonym":"Najdi Arabic",
+    "family":"Afro-Asiatic",
+    "flores_path":"ars_Arab",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"rng",
+    "speakers":1023339,
+    "language_name":"Ronga",
+    "autonym":"Ronga",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"luz",
+    "speakers":1019080,
+    "language_name":"Southern Luri",
+    "autonym":"Southern Luri",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"new",
+    "speakers":1000821,
+    "language_name":"Newari",
+    "autonym":"Newari",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"new",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tly",
+    "speakers":1000168,
+    "language_name":"Talysh",
+    "autonym":"Talysh",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"yi",
+    "speakers":997214,
+    "language_name":"Yiddish",
+    "autonym":"ייִדיש",
+    "family":"Indo-European",
+    "flores_path":"ydd_Hebr",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.5,
+    "commonvoice_locale":"yi",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"bez",
+    "speakers":995398,
+    "language_name":"Bena",
+    "autonym":"Hibena",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ksb",
+    "speakers":995398,
+    "language_name":"Shambala",
+    "autonym":"Kishambaa",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kfr",
+    "speakers":994568,
+    "language_name":"Kachhi",
+    "autonym":"Kachhi",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nij",
+    "speakers":987996,
+    "language_name":"Ngaju",
+    "autonym":"Ngaju",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mwk",
+    "speakers":977670,
+    "language_name":"Kita Maninkakan",
+    "autonym":"Kita Maninkakan",
+    "family":"Mande",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"jml",
+    "speakers":970493,
+    "language_name":"Jumli",
+    "autonym":"Jumli",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ryu",
+    "speakers":966404,
+    "language_name":"Central Okinawan",
+    "autonym":"Central Okinawan",
+    "family":"Japonic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mnw",
+    "speakers":966114,
+    "language_name":"Mon",
+    "autonym":"Mon",
+    "family":"Austroasiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"qug",
+    "speakers":963579,
+    "language_name":"Chimborazo Highland Quichua",
+    "autonym":"Chimborazo Highland Quichua",
+    "family":"Quechuan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"qug",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kac",
+    "speakers":962032,
+    "language_name":"Kachin",
+    "autonym":"Kachin",
+    "family":"Sino-Tibetan",
+    "flores_path":"kac_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"hnd",
+    "speakers":957354,
+    "language_name":"Southern Hindko",
+    "autonym":"Southern Hindko",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"li",
+    "speakers":950422,
+    "language_name":"Limburgish",
+    "autonym":"Limburgish",
+    "family":"Indo-European",
+    "flores_path":"lim_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"mxc",
+    "speakers":945510,
+    "language_name":"Manyika",
+    "autonym":"Manyika",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ce",
+    "speakers":935365,
+    "language_name":"Chechen",
+    "autonym":"Нохчийн",
+    "family":"Nakh-Daghestanian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kj",
+    "speakers":920524,
+    "language_name":"Kuanyama",
+    "autonym":"Kuanyama",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nr",
+    "speakers":903418,
+    "language_name":"South Ndebele",
+    "autonym":"South Ndebele",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"nr",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"cy",
+    "speakers":884910,
+    "language_name":"Welsh",
+    "autonym":"Cymraeg",
+    "family":"Indo-European",
+    "flores_path":"cym_Latn",
+    "fleurs_tag":"cy_gb",
+    "commonvoice_hours":124.0,
+    "commonvoice_locale":"cy",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"ada",
+    "speakers":880206,
+    "language_name":"Adangme",
+    "autonym":"Adangme",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"et",
+    "speakers":878449,
+    "language_name":"Estonian",
+    "autonym":"Eesti",
+    "family":"Uralic",
+    "flores_path":"ekk_Latn",
+    "fleurs_tag":"et_ee",
+    "commonvoice_hours":58.0,
+    "commonvoice_locale":"et",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"prd",
+    "speakers":864342,
+    "language_name":"Parsi-Dari",
+    "autonym":"Parsi-Dari",
+    "family":"Bookkeeping",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"yua",
+    "speakers":861955,
+    "language_name":"Yucateco",
+    "autonym":"Yucateco",
+    "family":"Mayan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ewo",
+    "speakers":860095,
+    "language_name":"Ewondo",
+    "autonym":"Ewondo",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":19.0,
+    "commonvoice_locale":"ewo",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kge",
+    "speakers":854483,
+    "language_name":"Komering",
+    "autonym":"Komering",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ttj",
+    "speakers":821807,
+    "language_name":"Tooro",
+    "autonym":"Tooro",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ttj",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"grt",
+    "speakers":821563,
+    "language_name":"Garo",
+    "autonym":"Garo",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gaa",
+    "speakers":821526,
+    "language_name":"Ga",
+    "autonym":"Gã",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mgy",
+    "speakers":819739,
+    "language_name":"Mbunga",
+    "autonym":"Mbunga",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tet",
+    "speakers":816395,
+    "language_name":"Tetum",
+    "autonym":"Tetum",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"iba",
+    "speakers":816302,
+    "language_name":"Iban",
+    "autonym":"Iban",
+    "family":"Bookkeeping",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"iba",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ebu",
+    "speakers":802918,
+    "language_name":"Embu",
+    "autonym":"Kĩembu",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"swg",
+    "speakers":801597,
+    "language_name":"Swabian",
+    "autonym":"Swabian",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"hnj",
+    "speakers":781687,
+    "language_name":"Hmong Njua",
+    "autonym":"𞄀𞄄𞄰𞄩𞄍𞄜𞄰",
+    "family":"Hmong-Mien",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kck",
+    "speakers":770954,
+    "language_name":"Kalanga",
+    "autonym":"Kalanga",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"dty",
+    "speakers":758198,
+    "language_name":"Dotyali",
+    "autonym":"Dotyali",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"pcd",
+    "speakers":746330,
+    "language_name":"Picard",
+    "autonym":"Picard",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"fy",
+    "speakers":743057,
+    "language_name":"Western Frisian",
+    "autonym":"Frysk",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":66.0,
+    "commonvoice_locale":"fy-NL",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ccp",
+    "speakers":729137,
+    "language_name":"Chakma",
+    "autonym":"𑄌𑄋𑄴𑄟𑄳𑄦",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"yao",
+    "speakers":722357,
+    "language_name":"Yao",
+    "autonym":"Yao",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"aoz",
+    "speakers":720970,
+    "language_name":"Uab Meto",
+    "autonym":"Uab Meto",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"cps",
+    "speakers":720595,
+    "language_name":"Capiznon",
+    "autonym":"Capiznon",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"asa",
+    "speakers":702634,
+    "language_name":"Asu",
+    "autonym":"Kipare",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"blt",
+    "speakers":681177,
+    "language_name":"Tai Dam",
+    "autonym":"ꪼꪕꪒꪾ",
+    "family":"Tai-Kadai",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"wa",
+    "speakers":679801,
+    "language_name":"Walloon",
+    "autonym":"Walon",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ses",
+    "speakers":664816,
+    "language_name":"Koyraboro Senni",
+    "autonym":"Koyraboro Senni",
+    "family":"Songhay",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bfy",
+    "speakers":654424,
+    "language_name":"Bagheli",
+    "autonym":"Bagheli",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ast",
+    "speakers":650205,
+    "language_name":"Asturian",
+    "autonym":"Asturianu",
+    "family":"Indo-European",
+    "flores_path":"ast_Latn",
+    "fleurs_tag":"ast_es",
+    "commonvoice_hours":0.8,
+    "commonvoice_locale":"ast",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"lki",
+    "speakers":645417,
+    "language_name":"Laki",
+    "autonym":"Laki",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"unx",
+    "speakers":636523,
+    "language_name":"Munda",
+    "autonym":"Munda",
+    "family":"Bookkeeping",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lis",
+    "speakers":627309,
+    "language_name":"Lisu",
+    "autonym":"Lisu",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nqo",
+    "speakers":626370,
+    "language_name":"N’Ko",
+    "autonym":"ߒߞߏ",
+    "family":"Artificial Language",
+    "flores_path":"nqo_Nkoo",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"nqo",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"gos",
+    "speakers":622094,
+    "language_name":"Gronings",
+    "autonym":"Gronings",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"gos",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lol",
+    "speakers":620858,
+    "language_name":"Mongo",
+    "autonym":"Mongo",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"thl",
+    "speakers":606558,
+    "language_name":"Dangaura Tharu",
+    "autonym":"Dangaura Tharu",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nap",
+    "speakers":605306,
+    "language_name":"Neapolitan",
+    "autonym":"Neapolitan",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nus",
+    "speakers":591427,
+    "language_name":"Nuer",
+    "autonym":"Thok Nath",
+    "family":"Nilotic",
+    "flores_path":"nus_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"br",
+    "speakers":563140,
+    "language_name":"Breton",
+    "autonym":"Brezhoneg",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":29.0,
+    "commonvoice_locale":"br",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"pis",
+    "speakers":561780,
+    "language_name":"Pijin",
+    "autonym":"Pijin",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rcf",
+    "speakers":559185,
+    "language_name":"Réunion Creole French",
+    "autonym":"Réunion Creole French",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"av",
+    "speakers":552716,
+    "language_name":"Avaric",
+    "autonym":"Avaric",
+    "family":"Nakh-Daghestanian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ng",
+    "speakers":552315,
+    "language_name":"Ndonga",
+    "autonym":"Ndonga",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"os",
+    "speakers":541444,
+    "language_name":"Ossetic",
+    "autonym":"Ирон",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":1.0,
+    "commonvoice_locale":"os",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"udm",
+    "speakers":538544,
+    "language_name":"Udmurt",
+    "autonym":"Udmurt",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"udm",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lij",
+    "speakers":536663,
+    "language_name":"Ligurian",
+    "autonym":"Ligure",
+    "family":"Indo-European",
+    "flores_path":"lij_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":5.1,
+    "commonvoice_locale":"lij",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"kea",
+    "speakers":530762,
+    "language_name":"Kabuverdianu",
+    "autonym":"Kabuverdianu",
+    "family":"Indo-European",
+    "flores_path":"kea_Latn",
+    "fleurs_tag":"kea_cv",
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"rue",
+    "speakers":527075,
+    "language_name":"Rusyn",
+    "autonym":"Rusyn",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"chm",
+    "speakers":524371,
+    "language_name":"Mari",
+    "autonym":"Mari",
+    "family":"Uralic",
+    "flores_path":"mhr_Cyrl",
+    "fleurs_tag":null,
+    "commonvoice_hours":282.0,
+    "commonvoice_locale":"mhr",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"scn",
+    "speakers":511702,
+    "language_name":"Sicilian",
+    "autonym":"Sicilianu",
+    "family":"Indo-European",
+    "flores_path":"scn_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"scn",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"lag",
+    "speakers":509409,
+    "language_name":"Langi",
+    "autonym":"Kɨlaangi",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bft",
+    "speakers":502520,
+    "language_name":"Balti",
+    "autonym":"Balti",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":18.0,
+    "commonvoice_locale":"bft",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nhe",
+    "speakers":501735,
+    "language_name":"Eastern Huasteca Nahuatl",
+    "autonym":"Eastern Huasteca Nahuatl",
+    "family":"Uto-Aztecan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"nhe",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nhw",
+    "speakers":501735,
+    "language_name":"Western Huasteca Nahuatl",
+    "autonym":"Western Huasteca Nahuatl",
+    "family":"Uto-Aztecan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"szl",
+    "speakers":497670,
+    "language_name":"Silesian",
+    "autonym":"Ślōnski",
+    "family":"Indo-European",
+    "flores_path":"szl_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"kaa",
+    "speakers":489046,
+    "language_name":"Kara-Kalpak",
+    "autonym":"Kara-Kalpak",
+    "family":"Turkic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"kaa",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gju",
+    "speakers":467002,
+    "language_name":"Gujari",
+    "autonym":"Gujari",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":11.0,
+    "commonvoice_locale":"gju",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"srx",
+    "speakers":464132,
+    "language_name":"Sirmauri",
+    "autonym":"Sirmauri",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mt",
+    "speakers":457267,
+    "language_name":"Maltese",
+    "autonym":"Malti",
+    "family":"Afro-Asiatic",
+    "flores_path":"mlt_Latn",
+    "fleurs_tag":"mt_mt",
+    "commonvoice_hours":8.7,
+    "commonvoice_locale":"mt",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"bap",
+    "speakers":454918,
+    "language_name":"Bantawa",
+    "autonym":"Bantawa",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sah",
+    "speakers":453510,
+    "language_name":"Yakut",
+    "autonym":"Саха Тыла",
+    "family":"Turkic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":14.0,
+    "commonvoice_locale":"sah",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mls",
+    "speakers":451060,
+    "language_name":"Masalit",
+    "autonym":"Masalit",
+    "family":"Maban",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kaj",
+    "speakers":449459,
+    "language_name":"Jju",
+    "autonym":"Kaje",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ady",
+    "speakers":444583,
+    "language_name":"Adyghe",
+    "autonym":"Adyghe",
+    "family":"Abkhaz-Adyge",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":13.0,
+    "commonvoice_locale":"ady",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ybb",
+    "speakers":443920,
+    "language_name":"Yemba",
+    "autonym":"Yemba",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"xmf",
+    "speakers":439670,
+    "language_name":"Mingrelian",
+    "autonym":"Mingrelian",
+    "family":"Kartvelian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":11.0,
+    "commonvoice_locale":"xmf",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"myv",
+    "speakers":439338,
+    "language_name":"Erzya",
+    "autonym":"Эрзянь Кель",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":3.8,
+    "commonvoice_locale":"myv",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"dav",
+    "speakers":438929,
+    "language_name":"Taita",
+    "autonym":"Kitaita",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":9.3,
+    "commonvoice_locale":"dav",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"maz",
+    "speakers":437410,
+    "language_name":"Central Mazahua",
+    "autonym":"Central Mazahua",
+    "family":"Otomanguean",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"vun",
+    "speakers":433291,
+    "language_name":"Vunjo",
+    "autonym":"Kyivunjo",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rof",
+    "speakers":433291,
+    "language_name":"Rombo",
+    "autonym":"Kihorombo",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"rof",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"jmc",
+    "speakers":433291,
+    "language_name":"Machame",
+    "autonym":"Kimachame",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kjg",
+    "speakers":431949,
+    "language_name":"Khmu",
+    "autonym":"Khmu",
+    "family":"Austroasiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"fan",
+    "speakers":426451,
+    "language_name":"Fang",
+    "autonym":"Fang",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":9.3,
+    "commonvoice_locale":"fan",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"krj",
+    "speakers":425806,
+    "language_name":"Kinaray-a",
+    "autonym":"Kinaray-A",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kdt",
+    "speakers":421207,
+    "language_name":"Kuy",
+    "autonym":"Kuy",
+    "family":"Austroasiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lb",
+    "speakers":421015,
+    "language_name":"Luxembourgish",
+    "autonym":"Lëtzebuergesch",
+    "family":"Indo-European",
+    "flores_path":"ltz_Latn",
+    "fleurs_tag":"lb_lu",
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"lb",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"srn",
+    "speakers":414507,
+    "language_name":"Sranan Tongo",
+    "autonym":"Sranan Tongo",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"dyo",
+    "speakers":409146,
+    "language_name":"Jola-Fonyi",
+    "autonym":"Joola",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tdg",
+    "speakers":394263,
+    "language_name":"Western Tamang",
+    "autonym":"Western Tamang",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"pnt",
+    "speakers":392463,
+    "language_name":"Pontic",
+    "autonym":"Pontic",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"zmi",
+    "speakers":391825,
+    "language_name":"Negeri Sembilan Malay",
+    "autonym":"Negeri Sembilan Malay",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nnh",
+    "speakers":388430,
+    "language_name":"Ngiemboon",
+    "autonym":"Shwóŋò Ngiembɔɔn",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":23.0,
+    "commonvoice_locale":"nnh",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bbj",
+    "speakers":388430,
+    "language_name":"Ghomala",
+    "autonym":"Ghomala",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":13.0,
+    "commonvoice_locale":"bbj",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"dv",
+    "speakers":388044,
+    "language_name":"Divehi",
+    "autonym":"Divehi",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":39.0,
+    "commonvoice_locale":"dv",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"saz",
+    "speakers":384566,
+    "language_name":"Saurashtra",
+    "autonym":"Saurashtra",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"hif",
+    "speakers":383749,
+    "language_name":"Fiji Hindi",
+    "autonym":"Fiji Hindi",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"fia",
+    "speakers":378161,
+    "language_name":"Nobiin",
+    "autonym":"Nobiin",
+    "family":"Nubian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bgx",
+    "speakers":377280,
+    "language_name":"Balkan Gagauz Turkish",
+    "autonym":"Balkan Gagauz Turkish",
+    "family":"Turkic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kvr",
+    "speakers":373836,
+    "language_name":"Kerinci",
+    "autonym":"Kerinci",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kvx",
+    "speakers":373602,
+    "language_name":"Parkari Koli",
+    "autonym":"Parkari Koli",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":12.0,
+    "commonvoice_locale":"kvx",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"dz",
+    "speakers":370341,
+    "language_name":"Dzongkha",
+    "autonym":"རྫོང་ཁ",
+    "family":"Bookkeeping",
+    "flores_path":"dzo_Tibt",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"pko",
+    "speakers":369343,
+    "language_name":"Pökoot",
+    "autonym":"Pökoot",
+    "family":"Nilotic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"dar",
+    "speakers":368477,
+    "language_name":"Dargwa",
+    "autonym":"Dargwa",
+    "family":"Nakh-Daghestanian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"dar",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lif",
+    "speakers":368085,
+    "language_name":"Limbu",
+    "autonym":"Limbu",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"fj",
+    "speakers":365030,
+    "language_name":"Fijian",
+    "autonym":"Fijian",
+    "family":"Austronesian",
+    "flores_path":"fij_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"thr",
+    "speakers":363935,
+    "language_name":"Rana Tharu",
+    "autonym":"Rana Tharu",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bkm",
+    "speakers":360685,
+    "language_name":"Kom",
+    "autonym":"Kom",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":13.0,
+    "commonvoice_locale":"bkm",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"is",
+    "speakers":350734,
+    "language_name":"Icelandic",
+    "autonym":"Íslenska",
+    "family":"Indo-European",
+    "flores_path":"isl_Latn",
+    "fleurs_tag":"is_is",
+    "commonvoice_hours":0.1,
+    "commonvoice_locale":"is",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"khw",
+    "speakers":350252,
+    "language_name":"Khowar",
+    "autonym":"Khowar",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":18.0,
+    "commonvoice_locale":"khw",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lbw",
+    "speakers":347134,
+    "language_name":"Tolaki",
+    "autonym":"Tolaki",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nxq",
+    "speakers":334565,
+    "language_name":"Naxi",
+    "autonym":"Naxi",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mgp",
+    "speakers":333607,
+    "language_name":"Eastern Magar",
+    "autonym":"Eastern Magar",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bax",
+    "speakers":332940,
+    "language_name":"Bamun",
+    "autonym":"Bamun",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":11.0,
+    "commonvoice_locale":"bax",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bas",
+    "speakers":332940,
+    "language_name":"Basaa",
+    "autonym":"Ɓàsàa",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":12.0,
+    "commonvoice_locale":"bas",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"khq",
+    "speakers":332408,
+    "language_name":"Koyra Chiini",
+    "autonym":"Koyra Ciini",
+    "family":"Songhay",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mvy",
+    "speakers":326901,
+    "language_name":"Indus Kohistani",
+    "autonym":"Indus Kohistani",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":23.0,
+    "commonvoice_locale":"mvy",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gay",
+    "speakers":320431,
+    "language_name":"Gayo",
+    "autonym":"Gayo",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"zdj",
+    "speakers":313124,
+    "language_name":"Ngazidja Comorian",
+    "autonym":"Ngazidja Comorian",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bua",
+    "speakers":311788,
+    "language_name":"Buriat",
+    "autonym":"Buriat",
+    "family":"Mongolic-Khitan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"bxr",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bto",
+    "speakers":305707,
+    "language_name":"Rinconada Bikol",
+    "autonym":"Rinconada Bikol",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"byv",
+    "speakers":305195,
+    "language_name":"Medumba",
+    "autonym":"Medumba",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":14.0,
+    "commonvoice_locale":"byv",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"njo",
+    "speakers":305001,
+    "language_name":"Ao Naga",
+    "autonym":"Ao Naga",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bfq",
+    "speakers":305001,
+    "language_name":"Badaga",
+    "autonym":"Badaga",
+    "family":"Dravidian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"thq",
+    "speakers":303279,
+    "language_name":"Kochila Tharu",
+    "autonym":"Kochila Tharu",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mdf",
+    "speakers":297616,
+    "language_name":"Moksha",
+    "autonym":"Мокшень Кяль",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.5,
+    "commonvoice_locale":"mdf",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rob",
+    "speakers":293729,
+    "language_name":"Tae'",
+    "autonym":"Tae'",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nzi",
+    "speakers":293402,
+    "language_name":"Nzima",
+    "autonym":"Nzima",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"naq",
+    "speakers":289308,
+    "language_name":"Nama",
+    "autonym":"Khoekhoegowab",
+    "family":"Khoe-Kwadi",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"wni",
+    "speakers":287736,
+    "language_name":"Ndzwani Comorian",
+    "autonym":"Ndzwani Comorian",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kum",
+    "speakers":283444,
+    "language_name":"Kumyk",
+    "autonym":"Kumyk",
+    "family":"Turkic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mua",
+    "speakers":277450,
+    "language_name":"Mundang",
+    "autonym":"Mundaŋ",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":2.3,
+    "commonvoice_locale":"mua",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"arn",
+    "speakers":272802,
+    "language_name":"Mapuche",
+    "autonym":"Mapudungun",
+    "family":"Araucanian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"arn",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"cja",
+    "speakers":270832,
+    "language_name":"Western Cham",
+    "autonym":"Western Cham",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bi",
+    "speakers":268500,
+    "language_name":"Bislama",
+    "autonym":"Bislama",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"khb",
+    "speakers":264864,
+    "language_name":"Lü",
+    "autonym":"Lü",
+    "family":"Tai-Kadai",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tdd",
+    "speakers":264864,
+    "language_name":"Tai Nüa",
+    "autonym":"Tai Nüa",
+    "family":"Tai-Kadai",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kxp",
+    "speakers":256851,
+    "language_name":"Wadiyara Koli",
+    "autonym":"Wadiyara Koli",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":12.0,
+    "commonvoice_locale":"kxp",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gjk",
+    "speakers":256851,
+    "language_name":"Kachi Koli",
+    "autonym":"Kachi Koli",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":11.0,
+    "commonvoice_locale":"gjk",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lez",
+    "speakers":255100,
+    "language_name":"Lezghian",
+    "autonym":"Lezghian",
+    "family":"Nakh-Daghestanian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kv",
+    "speakers":255100,
+    "language_name":"Komi",
+    "autonym":"Komi",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"kpv",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sm",
+    "speakers":252717,
+    "language_name":"Samoan",
+    "autonym":"Samoan",
+    "family":"Austronesian",
+    "flores_path":"smo_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"mrd",
+    "speakers":251722,
+    "language_name":"Western Magar",
+    "autonym":"Western Magar",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"saq",
+    "speakers":246228,
+    "language_name":"Samburu",
+    "autonym":"Kisampur",
+    "family":"Nilotic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"crh",
+    "speakers":245968,
+    "language_name":"Crimean Tatar",
+    "autonym":"Crimean Tatar",
+    "family":"Turkic",
+    "flores_path":"crh_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"crh",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"mdr",
+    "speakers":245664,
+    "language_name":"Mandar",
+    "autonym":"Mandar",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sxn",
+    "speakers":245664,
+    "language_name":"Sangir",
+    "autonym":"Sangir",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ext",
+    "speakers":245077,
+    "language_name":"Extremaduran",
+    "autonym":"Extremaduran",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nch",
+    "speakers":244435,
+    "language_name":"Central Huasteca Nahuatl",
+    "autonym":"Central Huasteca Nahuatl",
+    "family":"Uto-Aztecan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"zea",
+    "speakers":241926,
+    "language_name":"Zeelandic",
+    "autonym":"Zeelandic",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"krc",
+    "speakers":240927,
+    "language_name":"Karachay-Balkar",
+    "autonym":"Karachay-Balkar",
+    "family":"Turkic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"krc",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ksh",
+    "speakers":240479,
+    "language_name":"Colognian",
+    "autonym":"Kölsch",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"hz",
+    "speakers":239336,
+    "language_name":"Herero",
+    "autonym":"Herero",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sav",
+    "speakers":236046,
+    "language_name":"Saafi-Saafi",
+    "autonym":"Saafi-Saafi",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"zag",
+    "speakers":232364,
+    "language_name":"Zaghawa",
+    "autonym":"Zaghawa",
+    "family":"Saharan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"inh",
+    "speakers":226755,
+    "language_name":"Ingush",
+    "autonym":"Ingush",
+    "family":"Nakh-Daghestanian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ssy",
+    "speakers":218923,
+    "language_name":"Saho",
+    "autonym":"Saho",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"dtm",
+    "speakers":215087,
+    "language_name":"Tomo Kan Dogon",
+    "autonym":"Tomo Kan Dogon",
+    "family":"Dogon",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"pap",
+    "speakers":211640,
+    "language_name":"Papiamento",
+    "autonym":"Papiamentu",
+    "family":"Indo-European",
+    "flores_path":"pap_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"pap-AW",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"syr",
+    "speakers":210659,
+    "language_name":"Syriac",
+    "autonym":"ܣܘܪܝܝܐ",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"syr",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tog",
+    "speakers":207727,
+    "language_name":"Nyasa Tonga",
+    "autonym":"Nyasa Tonga",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"maf",
+    "speakers":205313,
+    "language_name":"Mafa",
+    "autonym":"Mafa",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"puu",
+    "speakers":200782,
+    "language_name":"Punu",
+    "autonym":"Punu",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kcg",
+    "speakers":199046,
+    "language_name":"Tyap",
+    "autonym":"Katab",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kao",
+    "speakers":195534,
+    "language_name":"Xaasongaxango",
+    "autonym":"Xaasongaxango",
+    "family":"Mande",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tyv",
+    "speakers":184239,
+    "language_name":"Tuvinian",
+    "autonym":"Tuvinian",
+    "family":"Turkic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"tyv",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"dtp",
+    "speakers":182852,
+    "language_name":"Central Dusun",
+    "autonym":"Central Dusun",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ria",
+    "speakers":172392,
+    "language_name":"Riang (India)",
+    "autonym":"Riang (India)",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"swb",
+    "speakers":170720,
+    "language_name":"Comorian",
+    "autonym":"Comorian",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bmq",
+    "speakers":168159,
+    "language_name":"Bomu",
+    "autonym":"Bomu",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ltg",
+    "speakers":167429,
+    "language_name":"Latgalian",
+    "autonym":"Latgalian",
+    "family":"Indo-European",
+    "flores_path":"ltg_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":29.0,
+    "commonvoice_locale":"ltg",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"nv",
+    "speakers":166320,
+    "language_name":"Navajo",
+    "autonym":"Diné Bizaad",
+    "family":"Athabaskan-Eyak-Tlingit",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bze",
+    "speakers":166204,
+    "language_name":"Jenaama Bozo",
+    "autonym":"Jenaama Bozo",
+    "family":"Mande",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"co",
+    "speakers":162836,
+    "language_name":"Corsican",
+    "autonym":"Corsu",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"co",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bfd",
+    "speakers":158146,
+    "language_name":"Bafut",
+    "autonym":"Bafut",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":11.0,
+    "commonvoice_locale":"bfd",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"xsr",
+    "speakers":157705,
+    "language_name":"Sherpa",
+    "autonym":"Sherpa",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kiu",
+    "speakers":155833,
+    "language_name":"Kirmanjki",
+    "autonym":"Kirmanjki",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ho",
+    "speakers":152449,
+    "language_name":"Hiri Motu",
+    "autonym":"Hiri Motu",
+    "family":"Pidgin",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kkj",
+    "speakers":149823,
+    "language_name":"Kako",
+    "autonym":"Kakɔ",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bss",
+    "speakers":149823,
+    "language_name":"Akoose",
+    "autonym":"Akoose",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sly",
+    "speakers":144194,
+    "language_name":"Selayar",
+    "autonym":"Selayar",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mi",
+    "speakers":137913,
+    "language_name":"Māori",
+    "autonym":"Māori",
+    "family":"Austronesian",
+    "flores_path":"mri_Latn",
+    "fleurs_tag":"mi_nz",
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"rjs",
+    "speakers":133443,
+    "language_name":"Rajbanshi",
+    "autonym":"Rajbanshi",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"dua",
+    "speakers":133176,
+    "language_name":"Duala",
+    "autonym":"Duálá",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":14.0,
+    "commonvoice_locale":"dua",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"guc",
+    "speakers":132529,
+    "language_name":"Wayuu",
+    "autonym":"Wayuu",
+    "family":"Arawakan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"guc",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"vai",
+    "speakers":131906,
+    "language_name":"Vai",
+    "autonym":"ꕙꔤ",
+    "family":"Mande",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"taj",
+    "speakers":130410,
+    "language_name":"Eastern Tamang",
+    "autonym":"Eastern Tamang",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mgo",
+    "speakers":130401,
+    "language_name":"Metaʼ",
+    "autonym":"Metaʼ",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"pdc",
+    "speakers":129729,
+    "language_name":"Pennsylvania German",
+    "autonym":"Pennsylvania German",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rwk",
+    "speakers":128816,
+    "language_name":"Rwa",
+    "autonym":"Kiruwa",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"trw",
+    "speakers":123756,
+    "language_name":"Torwali",
+    "autonym":"توروالی",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":19.0,
+    "commonvoice_locale":"trw",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mfv",
+    "speakers":121170,
+    "language_name":"Mandjak",
+    "autonym":"Mandjak",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tsj",
+    "speakers":117348,
+    "language_name":"Tshangla",
+    "autonym":"Tshangla",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sbp",
+    "speakers":117106,
+    "language_name":"Sangu",
+    "autonym":"Ishisangu",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"krl",
+    "speakers":116212,
+    "language_name":"Karelian",
+    "autonym":"Karelian",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lad",
+    "speakers":112781,
+    "language_name":"Ladino",
+    "autonym":"Ladino",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ife",
+    "speakers":111910,
+    "language_name":"Ifè",
+    "autonym":"Ifè",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gag",
+    "speakers":111028,
+    "language_name":"Gagauz",
+    "autonym":"Gagauz",
+    "family":"Turkic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lbe",
+    "speakers":110543,
+    "language_name":"Lak",
+    "autonym":"Lak",
+    "family":"Nakh-Daghestanian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sdc",
+    "speakers":106085,
+    "language_name":"Sassarese Sardinian",
+    "autonym":"Sassarese Sardinian",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"to",
+    "speakers":100790,
+    "language_name":"Tongan",
+    "autonym":"Lea Fakatonga",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bjt",
+    "speakers":95992,
+    "language_name":"Balanta-Ganja",
+    "autonym":"Balanta-Ganja",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"jgo",
+    "speakers":94333,
+    "language_name":"Ngomba",
+    "autonym":"Ndaꞌa",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":3.9,
+    "commonvoice_locale":"jgo",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"crs",
+    "speakers":94061,
+    "language_name":"Seselwa Creole French",
+    "autonym":"Seselwa Creole French",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ab",
+    "speakers":91953,
+    "language_name":"Abkhazian",
+    "autonym":"Аԥсшәа",
+    "family":"Abkhaz-Adyge",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":68.0,
+    "commonvoice_locale":"ab",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ty",
+    "speakers":91488,
+    "language_name":"Tahitian",
+    "autonym":"Tahitian",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ty",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"iu",
+    "speakers":90466,
+    "language_name":"Inuktitut",
+    "autonym":"Inuktitut",
+    "family":"Eskimo-Aleut",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"pdt",
+    "speakers":90466,
+    "language_name":"Plautdietsch",
+    "autonym":"Plautdietsch",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bpy",
+    "speakers":90174,
+    "language_name":"Bishnupriya",
+    "autonym":"Bishnupriya",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ksf",
+    "speakers":88784,
+    "language_name":"Bafia",
+    "autonym":"Rikpa",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":20.0,
+    "commonvoice_locale":"ksf",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gvr",
+    "speakers":87951,
+    "language_name":"Gurung",
+    "autonym":"Gurung",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"cjm",
+    "speakers":87862,
+    "language_name":"Eastern Cham",
+    "autonym":"Eastern Cham",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lcp",
+    "speakers":87751,
+    "language_name":"Western Lawa",
+    "autonym":"Western Lawa",
+    "family":"Austroasiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"knf",
+    "speakers":83151,
+    "language_name":"Mankanya",
+    "autonym":"Mankanya",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lep",
+    "speakers":79743,
+    "language_name":"Lepcha",
+    "autonym":"Lepcha",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"byn",
+    "speakers":79056,
+    "language_name":"Blin",
+    "autonym":"Blin",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"btv",
+    "speakers":78843,
+    "language_name":"Bateri",
+    "autonym":"Bateri",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.6,
+    "commonvoice_locale":"btv",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tkt",
+    "speakers":72787,
+    "language_name":"Kathoriya Tharu",
+    "autonym":"Kathoriya Tharu",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gd",
+    "speakers":72337,
+    "language_name":"Scottish Gaelic",
+    "autonym":"Gàidhlig",
+    "family":"Indo-European",
+    "flores_path":"gla_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"fo",
+    "speakers":71351,
+    "language_name":"Faroese",
+    "autonym":"Føroyskt",
+    "family":"Indo-European",
+    "flores_path":"fao_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"fo",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"vro",
+    "speakers":70031,
+    "language_name":"Võro",
+    "autonym":"Võro",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ken",
+    "speakers":69362,
+    "language_name":"Kenyang",
+    "autonym":"Kɛnyaŋ",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gil",
+    "speakers":67078,
+    "language_name":"Gilbertese",
+    "autonym":"Gilbertese",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bvb",
+    "speakers":66058,
+    "language_name":"Bube",
+    "autonym":"Bube",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mwv",
+    "speakers":64086,
+    "language_name":"Mentawai",
+    "autonym":"Mentawai",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"frp",
+    "speakers":63777,
+    "language_name":"Arpitan",
+    "autonym":"Arpitan",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"koi",
+    "speakers":63775,
+    "language_name":"Komi-Permyak",
+    "autonym":"Komi-Permyak",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kfo",
+    "speakers":63207,
+    "language_name":"Koro",
+    "autonym":"Koro",
+    "family":"Mande",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mh",
+    "speakers":56879,
+    "language_name":"Marshallese",
+    "autonym":"Marshallese",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"fit",
+    "speakers":56114,
+    "language_name":"Tornedalen Finnish",
+    "autonym":"Tornedalen Finnish",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kl",
+    "speakers":55440,
+    "language_name":"Kalaallisut",
+    "autonym":"Kalaallisut",
+    "family":"Eskimo-Aleut",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bra",
+    "speakers":54370,
+    "language_name":"Braj",
+    "autonym":"Braj",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gcr",
+    "speakers":51872,
+    "language_name":"Guianese Creole French",
+    "autonym":"Guianese Creole French",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"se",
+    "speakers":51530,
+    "language_name":"Northern Sami",
+    "autonym":"Davvisámegiella",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"blo",
+    "speakers":51507,
+    "language_name":"Anii",
+    "autonym":"Anii Kagɩja",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kgp",
+    "speakers":50812,
+    "language_name":"Kaingang",
+    "autonym":"Kanhgág",
+    "family":"Nuclear-Macro-Je",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"csb",
+    "speakers":49767,
+    "language_name":"Kashubian",
+    "autonym":"Kashubian",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bqv",
+    "speakers":46718,
+    "language_name":"Koro Wachi",
+    "autonym":"Koro Wachi",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ch",
+    "speakers":46325,
+    "language_name":"Chamorro",
+    "autonym":"Chamorro",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"cch",
+    "speakers":44946,
+    "language_name":"Atsam",
+    "autonym":"Atsam",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"buc",
+    "speakers":44620,
+    "language_name":"Bushi",
+    "autonym":"Bushi",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rm",
+    "speakers":42020,
+    "language_name":"Romansh",
+    "autonym":"Rumantsch",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"agq",
+    "speakers":38843,
+    "language_name":"Aghem",
+    "autonym":"Aghem",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kxv",
+    "speakers":38457,
+    "language_name":"Kuvi",
+    "autonym":"Kuvi",
+    "family":"Dravidian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"snf",
+    "speakers":37767,
+    "language_name":"Noon",
+    "autonym":"Noon",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"fur",
+    "speakers":37442,
+    "language_name":"Friulian",
+    "autonym":"Furlan",
+    "family":"Indo-European",
+    "flores_path":"fur_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"tdh",
+    "speakers":36393,
+    "language_name":"Thulung",
+    "autonym":"Thulung",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"egl",
+    "speakers":31201,
+    "language_name":"Emilian",
+    "autonym":"Emilian",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"chk",
+    "speakers":30731,
+    "language_name":"Chuukese",
+    "autonym":"Chuukese",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mrj",
+    "speakers":29762,
+    "language_name":"Western Mari",
+    "autonym":"Western Mari",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":34.0,
+    "commonvoice_locale":"mrj",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"haw",
+    "speakers":29605,
+    "language_name":"Hawaiian",
+    "autonym":"ʻŌlelo HawaiʻI",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mro",
+    "speakers":29277,
+    "language_name":"Mru",
+    "autonym":"Mru",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"frc",
+    "speakers":27942,
+    "language_name":"Cajun French",
+    "autonym":"Cajun French",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"yrl",
+    "speakers":26171,
+    "language_name":"Nheengatu",
+    "autonym":"Nheẽgatu",
+    "family":"Tupian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"an",
+    "speakers":26008,
+    "language_name":"Aragonese",
+    "autonym":"Aragonés",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":17.0,
+    "commonvoice_locale":"an",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"chr",
+    "speakers":25613,
+    "language_name":"Cherokee",
+    "autonym":"Ꮳꮃꭹ",
+    "family":"Iroquoian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rmo",
+    "speakers":24372,
+    "language_name":"Sinte Romani",
+    "autonym":"Sinte Romani",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"oj",
+    "speakers":23747,
+    "language_name":"Ojibwa",
+    "autonym":"Ojibwa",
+    "family":"Algic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"pon",
+    "speakers":23560,
+    "language_name":"Pohnpeian",
+    "autonym":"Pohnpeian",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lzz",
+    "speakers":22965,
+    "language_name":"Laz",
+    "autonym":"Laz",
+    "family":"Kartvelian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"lzz",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ttt",
+    "speakers":22453,
+    "language_name":"Muslim Tat",
+    "autonym":"Muslim Tat",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"esu",
+    "speakers":20956,
+    "language_name":"Central Yupik",
+    "autonym":"Central Yupik",
+    "family":"Eskimo-Aleut",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":7.6,
+    "commonvoice_locale":"esu",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"dak",
+    "speakers":20832,
+    "language_name":"Dakota",
+    "autonym":"Dakota",
+    "family":"Siouan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"alt",
+    "speakers":19841,
+    "language_name":"Southern Altai",
+    "autonym":"Southern Altai",
+    "family":"Turkic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"amo",
+    "speakers":18620,
+    "language_name":"Amo",
+    "autonym":"Amo",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gub",
+    "speakers":17784,
+    "language_name":"Guajajára",
+    "autonym":"Guajajára",
+    "family":"Tupian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"hnn",
+    "speakers":17469,
+    "language_name":"Hanunoo",
+    "autonym":"Hanunoo",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tkr",
+    "speakers":16329,
+    "language_name":"Tsakhur",
+    "autonym":"Tsakhur",
+    "family":"Nakh-Daghestanian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"pau",
+    "speakers":16047,
+    "language_name":"Palauan",
+    "autonym":"Palauan",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sa",
+    "speakers":15913,
+    "language_name":"Sanskrit",
+    "autonym":"संस्कृत भाषा",
+    "family":"Indo-European",
+    "flores_path":"san_Deva",
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"bsc",
+    "speakers":15264,
+    "language_name":"Bassari",
+    "autonym":"Bassari",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ojs",
+    "speakers":15078,
+    "language_name":"Oji-Cree",
+    "autonym":"Oji-Cree",
+    "family":"Algic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kht",
+    "speakers":13527,
+    "language_name":"Khamti",
+    "autonym":"Khamti",
+    "family":"Tai-Kadai",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"hsb",
+    "speakers":12826,
+    "language_name":"Upper Sorbian",
+    "autonym":"Hornjoserbšćina",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":2.9,
+    "commonvoice_locale":"hsb",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"chp",
+    "speakers":12816,
+    "language_name":"Chipewyan",
+    "autonym":"Chipewyan",
+    "family":"Athabaskan-Eyak-Tlingit",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"moe",
+    "speakers":12062,
+    "language_name":"Innu-aimun",
+    "autonym":"Innu-Aimun",
+    "family":"Algic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sli",
+    "speakers":11868,
+    "language_name":"Lower Silesian",
+    "autonym":"Lower Silesian",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"wae",
+    "speakers":11377,
+    "language_name":"Walser",
+    "autonym":"Walser",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"cho",
+    "speakers":10977,
+    "language_name":"Choctaw",
+    "autonym":"Chahta",
+    "family":"Muskogean",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tbw",
+    "speakers":10045,
+    "language_name":"Tagbanwa",
+    "autonym":"Tagbanwa",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"xav",
+    "speakers":9951,
+    "language_name":"Xavánte",
+    "autonym":"Xavánte",
+    "family":"Nuclear-Macro-Je",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tvl",
+    "speakers":9868,
+    "language_name":"Tuvalu",
+    "autonym":"Tuvalu",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"frr",
+    "speakers":9619,
+    "language_name":"Northern Frisian",
+    "autonym":"Nordfriisk",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rug",
+    "speakers":9591,
+    "language_name":"Roviana",
+    "autonym":"Roviana",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"wls",
+    "speakers":9512,
+    "language_name":"Wallisian",
+    "autonym":"Wallisian",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rmu",
+    "speakers":9488,
+    "language_name":"Tavringer Romani",
+    "autonym":"Tavringer Romani",
+    "family":"Speech Register",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"cr",
+    "speakers":9047,
+    "language_name":"Cree",
+    "autonym":"Cree",
+    "family":"Algic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nmg",
+    "speakers":8878,
+    "language_name":"Kwasio",
+    "autonym":"Kwasio",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lkt",
+    "speakers":8316,
+    "language_name":"Lakota",
+    "autonym":"LakȟólʼIyapi",
+    "family":"Siouan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kos",
+    "speakers":7990,
+    "language_name":"Kosraean",
+    "autonym":"Kosraean",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ik",
+    "speakers":7983,
+    "language_name":"Inupiaq",
+    "autonym":"Inupiaq",
+    "family":"Eskimo-Aleut",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":7.2,
+    "commonvoice_locale":"ipk",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gbz",
+    "speakers":7983,
+    "language_name":"Zoroastrian Dari",
+    "autonym":"Zoroastrian Dari",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"twq",
+    "speakers":7970,
+    "language_name":"Tasawaq",
+    "autonym":"Tasawaq Senni",
+    "family":"Songhay",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bku",
+    "speakers":7970,
+    "language_name":"Buhid",
+    "autonym":"Buhid",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mic",
+    "speakers":7916,
+    "language_name":"Mi'kmaw",
+    "autonym":"LʼNuiʼSuti",
+    "family":"Algic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mey",
+    "speakers":7239,
+    "language_name":"Hassaniyya",
+    "autonym":"Hassaniyya",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"dsb",
+    "speakers":6974,
+    "language_name":"Lower Sorbian",
+    "autonym":"Dolnoserbšćina",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"dsb",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"na",
+    "speakers":6930,
+    "language_name":"Nauru",
+    "autonym":"Nauru",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lwl",
+    "speakers":6898,
+    "language_name":"Eastern Lawa",
+    "autonym":"Eastern Lawa",
+    "family":"Austroasiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"yap",
+    "speakers":6556,
+    "language_name":"Yapese",
+    "autonym":"Yapese",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"atj",
+    "speakers":6408,
+    "language_name":"Atikamekw",
+    "autonym":"Atikamekw",
+    "family":"Algic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"pms",
+    "speakers":6178,
+    "language_name":"Piedmontese",
+    "autonym":"Piedmontese",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rmf",
+    "speakers":5015,
+    "language_name":"Kalo Finnish Romani",
+    "autonym":"Kalo Finnish Romani",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"bla",
+    "speakers":4900,
+    "language_name":"Siksiká",
+    "autonym":"Siksiká",
+    "family":"Algic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"fud",
+    "speakers":4756,
+    "language_name":"East Futuna",
+    "autonym":"East Futuna",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"trv",
+    "speakers":4721,
+    "language_name":"Taroko",
+    "autonym":"Patas Taroko",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":10.0,
+    "commonvoice_locale":"trv",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"crk",
+    "speakers":4146,
+    "language_name":"Plains Cree",
+    "autonym":"Plains Cree",
+    "family":"Algic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"saf",
+    "speakers":4108,
+    "language_name":"Safaliba",
+    "autonym":"Safaliba",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"mus",
+    "speakers":3992,
+    "language_name":"Muscogee",
+    "autonym":"Mvskoke",
+    "family":"Muskogean",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"vep",
+    "speakers":3543,
+    "language_name":"Veps",
+    "autonym":"Veps",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tnr",
+    "speakers":3305,
+    "language_name":"Ménik",
+    "autonym":"Ménik",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"vic",
+    "speakers":3113,
+    "language_name":"Virgin Islands Creole English",
+    "autonym":"Virgin Islands Creole English",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tru",
+    "speakers":3035,
+    "language_name":"Turoyo",
+    "autonym":"Turoyo",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"uli",
+    "speakers":2971,
+    "language_name":"Ulithian",
+    "autonym":"Ulithian",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rtm",
+    "speakers":2527,
+    "language_name":"Rotuman",
+    "autonym":"Rotuman",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"wbp",
+    "speakers":2496,
+    "language_name":"Warlpiri",
+    "autonym":"Warlpiri",
+    "family":"Pama-Nyungan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"yav",
+    "speakers":2303,
+    "language_name":"Yangben",
+    "autonym":"Nuasue",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"yav",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"den",
+    "speakers":2299,
+    "language_name":"Slave",
+    "autonym":"Slave",
+    "family":"Athabaskan-Eyak-Tlingit",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"dgr",
+    "speakers":2111,
+    "language_name":"Dogrib",
+    "autonym":"Dogrib",
+    "family":"Athabaskan-Eyak-Tlingit",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"frs",
+    "speakers":2004,
+    "language_name":"Eastern Frisian",
+    "autonym":"Eastern Frisian",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kw",
+    "speakers":1973,
+    "language_name":"Cornish",
+    "autonym":"Kernewek",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":13.0,
+    "commonvoice_locale":"kw",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"csw",
+    "speakers":1809,
+    "language_name":"Swampy Cree",
+    "autonym":"ᓀᐦᐃᓇᐍᐏᐣ",
+    "family":"Algic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"moh",
+    "speakers":1772,
+    "language_name":"Mohawk",
+    "autonym":"KanienʼKéha",
+    "family":"Iroquoian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gv",
+    "speakers":1719,
+    "language_name":"Manx",
+    "autonym":"Gaelg",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":11.0,
+    "commonvoice_locale":"gv",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"smj",
+    "speakers":1530,
+    "language_name":"Lule Sami",
+    "autonym":"Julevsámegiella",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"nsk",
+    "speakers":1395,
+    "language_name":"Naskapi",
+    "autonym":"Naskapi",
+    "family":"Algic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tkl",
+    "speakers":1285,
+    "language_name":"Tokelau",
+    "autonym":"Tokelau",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"niu",
+    "speakers":1120,
+    "language_name":"Niuean",
+    "autonym":"Niuean",
+    "family":"Austronesian",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"stq",
+    "speakers":962,
+    "language_name":"Saterland Frisian",
+    "autonym":"Saterland Frisian",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sei",
+    "speakers":901,
+    "language_name":"Seri",
+    "autonym":"Seri",
+    "family":null,
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":11.0,
+    "commonvoice_locale":"sei",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"clc",
+    "speakers":867,
+    "language_name":"Chilcotin",
+    "autonym":"Chilcotin",
+    "family":"Athabaskan-Eyak-Tlingit",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"la",
+    "speakers":820,
+    "language_name":"Latin",
+    "autonym":"Lingua Latina",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"hur",
+    "speakers":716,
+    "language_name":"Halkomelem",
+    "autonym":"Halkomelem",
+    "family":"Salishan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"crg",
+    "speakers":678,
+    "language_name":"Michif",
+    "autonym":"Michif",
+    "family":"Algic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sms",
+    "speakers":613,
+    "language_name":"Skolt Sami",
+    "autonym":"SääʹMǩiõll",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"smn",
+    "speakers":613,
+    "language_name":"Inari Sami",
+    "autonym":"Anarâškielâ",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lil",
+    "speakers":528,
+    "language_name":"Lillooet",
+    "autonym":"Lillooet",
+    "family":"Salishan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"oka",
+    "speakers":490,
+    "language_name":"Okanagan",
+    "autonym":"Okanagan",
+    "family":"Salishan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"pqm",
+    "speakers":490,
+    "language_name":"Maliseet-Passamaquoddy",
+    "autonym":"Maliseet-Passamaquoddy",
+    "family":"Algic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"kwk",
+    "speakers":377,
+    "language_name":"Kwakʼwala",
+    "autonym":"KwakʼWala",
+    "family":"Wakashan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"crl",
+    "speakers":377,
+    "language_name":"Northern East Cree",
+    "autonym":"Northern East Cree",
+    "family":"Algic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gwi",
+    "speakers":302,
+    "language_name":"Gwichʼin",
+    "autonym":"GwichʼIn",
+    "family":"Athabaskan-Eyak-Tlingit",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"eo",
+    "speakers":301,
+    "language_name":"Esperanto",
+    "autonym":"Esperanto",
+    "family":"Artificial Language",
+    "flores_path":"epo_Latn",
+    "fleurs_tag":null,
+    "commonvoice_hours":1436.0,
+    "commonvoice_locale":"eo",
+    "in_benchmark":true
+  },
+  {
+    "bcp_47":"sma",
+    "speakers":296,
+    "language_name":"Southern Sami",
+    "autonym":"Åarjelsaemien Gïele",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"tsd",
+    "speakers":202,
+    "language_name":"Tsakonian",
+    "autonym":"Tsakonian",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"izh",
+    "speakers":142,
+    "language_name":"Ingrian",
+    "autonym":"Ingrian",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"izh",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ia",
+    "speakers":136,
+    "language_name":"Interlingua",
+    "autonym":"Interlingua",
+    "family":"Artificial Language",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":15.0,
+    "commonvoice_locale":"ia",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"aro",
+    "speakers":105,
+    "language_name":"Araona",
+    "autonym":"Araona",
+    "family":"Pano-Tacanan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"prg",
+    "speakers":38,
+    "language_name":"Prussian",
+    "autonym":"Prūsiskan",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ie",
+    "speakers":1,
+    "language_name":"Interlingue",
+    "autonym":"Interlingue",
+    "family":"Artificial Language",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"ie",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"lzh",
+    "speakers":0,
+    "language_name":"Literary Chinese",
+    "autonym":"Literary Chinese",
+    "family":"Sino-Tibetan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"io",
+    "speakers":0,
+    "language_name":"Ido",
+    "autonym":"Ido",
+    "family":"Artificial Language",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"jbo",
+    "speakers":0,
+    "language_name":"Lojban",
+    "autonym":"La .Lojban.",
+    "family":"Artificial Language",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.0,
+    "commonvoice_locale":"jbo",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"jut",
+    "speakers":0,
+    "language_name":"Jutish",
+    "autonym":"Jutish",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"vot",
+    "speakers":0,
+    "language_name":"Votic",
+    "autonym":"Votic",
+    "family":"Uralic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":0.1,
+    "commonvoice_locale":"vot",
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"gez",
+    "speakers":0,
+    "language_name":"Geez",
+    "autonym":"Geez",
+    "family":"Afro-Asiatic",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"osa",
+    "speakers":0,
+    "language_name":"Osage",
+    "autonym":"𐓏𐓘𐓻𐓘𐓻𐓟",
+    "family":"Siouan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"rgn",
+    "speakers":0,
+    "language_name":"Romagnol",
+    "autonym":"Romagnol",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"cu",
+    "speakers":0,
+    "language_name":"Church Slavic",
+    "autonym":"Church Slavic",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"sgs",
+    "speakers":0,
+    "language_name":"Samogitian",
+    "autonym":"Samogitian",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"ann",
+    "speakers":0,
+    "language_name":"Obolo",
+    "autonym":"Obolo",
+    "family":"Atlantic-Congo",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"cic",
+    "speakers":0,
+    "language_name":"Chickasaw",
+    "autonym":"Chikashshanompaʼ",
+    "family":"Muskogean",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"cad",
+    "speakers":0,
+    "language_name":"Caddo",
+    "autonym":"Caddo",
+    "family":"Caddoan",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  },
+  {
+    "bcp_47":"pfl",
+    "speakers":0,
+    "language_name":"Palatine German",
+    "autonym":"Palatine German",
+    "family":"Indo-European",
+    "flores_path":null,
+    "fleurs_tag":null,
+    "commonvoice_hours":null,
+    "commonvoice_locale":null,
+    "in_benchmark":false
+  }
+]
\ No newline at end of file