Improve audio handling compatibility

- Add support for different audio data formats
- Ensure proper .wav extension for Whisper compatibility 
- Add file size logging for debugging
- Fix data format conversion between client and server

app.py

@@ -429,6 +429,18 @@ def setup_realtime_interpreter():
                 
                 # Use whisper to transcribe
                 try:
+                    # Check if file is valid
+                    file_size = os.path.getsize(temp_path)
+                    logger.info(f"File size: {file_size} bytes")
+                    
+                    # Enforce .wav extension for compatibility
+                    new_temp_path = temp_path
+                    if not temp_path.lower().endswith('.wav'):
+                        new_temp_path = temp_path + '.wav'
+                        os.rename(temp_path, new_temp_path)
+                        logger.info(f"Renamed audio file to ensure .wav extension: {new_temp_path}")
+                        temp_path = new_temp_path
+                    
                     result = whisper_model.transcribe(
                         temp_path,
                         language=LANGUAGE_TO_CODE.get(source_lang, None)
@@ -626,6 +638,22 @@ def handle_audio_chunk(data):
         emit('error', {'message': 'No audio data received'})
         return
     
+    # Convert audio data to proper format if needed
+    if isinstance(audio_data, list):
+        # Convert list to bytes
+        audio_data = bytes(audio_data)
+        logger.info(f"Converted audio data from list to bytes: {len(audio_data)} bytes")
+    elif isinstance(audio_data, str):
+        # This might be base64 encoded data
+        try:
+            import base64
+            audio_data = base64.b64decode(audio_data.split(',')[1] if ',' in audio_data else audio_data)
+            logger.info(f"Decoded base64 audio data: {len(audio_data)} bytes")
+        except Exception as e:
+            logger.error(f"Error decoding audio data: {str(e)}")
+            emit('error', {'message': 'Invalid audio data format'})
+            return
+    
     # Get session details
     source_lang = active_sessions[request.sid]['source_lang']