fdaudens HF Staff commited on
Commit
27e0438
Β·
verified Β·
1 Parent(s): 3c3337a

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +7 -6
app.py CHANGED
@@ -184,12 +184,13 @@ def process_audio_rag(audio_file_path, query, chunk_length=30, use_openai=False,
184
  with gr.Blocks(title="AudioRAG Demo") as demo:
185
  gr.Markdown("# AudioRAG Demo - Semantic Audio Search")
186
  gr.Markdown("""
187
- This demo builds on the work from the ColQwen team, expanding retrieval capabilities beyond images to include audio and video. Inspired by the Qwen-Omni series, ColQwen-Omni (3B) pushes the boundaries of multimodal search β€” embedding and retrieving almost any type of content.
188
-
189
- **What’s new?**
190
- Unlike traditional methods, this model searches directly through raw audio without converting it to text. It understands semantic meaning in sound, speech, and audio patterns β€” making "AudioRAG" a real possibility.
191
 
192
- πŸ“– [Blog post](https://huggingface.co/blog/manu/colqwen-omni-omnimodal-retrieval) | πŸ€— [Model on Hugging Face](https://huggingface.co/vidore/colqwen-omni-v0.1)
 
 
193
  """)
194
 
195
  with gr.Row():
@@ -211,7 +212,7 @@ with gr.Blocks(title="AudioRAG Demo") as demo:
211
 
212
  gr.Examples(
213
  examples=[
214
- ["test.m4a", "Who's the guest of the podcast?", 30],
215
  ],
216
  inputs=[audio_input, query_input, chunk_length]
217
  )
 
184
  with gr.Blocks(title="AudioRAG Demo") as demo:
185
  gr.Markdown("# AudioRAG Demo - Semantic Audio Search")
186
  gr.Markdown("""
187
+ This demo builds on the work from the ColQwen team, expanding retrieval capabilities beyond images to include audio and video.
188
+
189
+ Unlike traditional methods, this model searches directly through raw audio without converting it to text. It understands semantic meaning in sound, speech, and audio patterns, making "AudioRAG" a real possibility.
 
190
 
191
+ πŸ“– [Blog post](https://huggingface.co/blog/manu/colqwen-omni-omnimodal-retrieval) | πŸ€— [Model on Hugging Face](https://huggingface.co/vidore/colqwen-omni-v0.1) | πŸ““ [Colab Notebook](https://colab.research.google.com/drive/1YOlTWfLbiyQqfq1SlqHA2iME1R-nH4aS#scrollTo=w7UyXtEcK0lA)
192
+
193
+ πŸŽ™οΈ Sample come from [Newsroom Robots](https://www.newsroomrobots.com/p/how-open-source-ai-puts-newsrooms)
194
  """)
195
 
196
  with gr.Row():
 
212
 
213
  gr.Examples(
214
  examples=[
215
+ ["test.m4a", "Who’s the podcast host?", 30],
216
  ],
217
  inputs=[audio_input, query_input, chunk_length]
218
  )