feat: Enable MCP

app.py

@@ -29,7 +29,21 @@ LANGUAGES = {
 
 @spaces.GPU()
 def process_audio(audio_path, model_name, lang_name, max_tokens=500):
-    """Process audio with selected Voxtral model and return the generated response"""
+    """Process audio with selected Voxtral model and return the generated response.
+    
+    This function takes an audio file and processes it using the selected Voxtral model
+    to generate a transcription in the specified language.
+    
+    Args:
+        audio_path: Path to the audio file to be transcribed.
+        model_name: Name of the Voxtral model to use ("Voxtral Mini (3B)" or "Voxtral Small (24B)").
+        lang_name: Name of the language for transcription (e.g., "English", "French", etc.).
+        max_tokens: Maximum number of tokens to generate in the output (default: 500).
+    
+    Returns:
+        String containing the transcribed text from the audio file, or an error message
+        if the audio file is missing or an invalid model is selected.
+    """
     if not audio_path:
         return "Please upload an audio file."
 
@@ -111,4 +125,4 @@ with gr.Blocks(title="Voxtral Demo") as demo:
 
 # Launch the app
 if __name__ == "__main__":
-    demo.queue().launch(share=False, ssr_mode=False)
+    demo.queue().launch(share=False, ssr_mode=False, mcp_server=True)