Update app.py

app.py

@@ -212,23 +212,37 @@ pipe_asr = pipeline(
 
 # Function to handle voice input, generate response from Neo4j, and return audio output
 def handle_voice_to_voice(audio):
-    # Transcribe audio input to text
-    sr, y = audio
-    
-    # Ensure that the audio is in float32 format
-    y = y.astype(np.float32)
-    y = y / np.max(np.abs(y))  # Normalize audio to range [-1.0, 1.0]
-
-    # Process the audio data with Whisper ASR
-    result = pipe_asr({"array": y, "sampling_rate": sr}, return_timestamps=False)
-    question = result.get("text", "")
-
-    # Get response using the transcribed question
-    response = get_response(question)
-
-    # Generate audio from the response
-    audio_path = generate_audio_elevenlabs(response)
-    return audio_path
+    try:
+        # Transcribe audio input to text
+        sr, y = audio
+        y = y.astype(np.float32)
+        y = y / np.max(np.abs(y))  # Normalize audio to range [-1.0, 1.0]
+
+        logging.debug(f"Audio sample rate: {sr}")
+        logging.debug(f"Audio data: {y[:100]}")  # Log first 100 samples for brevity
+
+        # Process the audio data with Whisper ASR
+        result = pipe_asr({"array": y, "sampling_rate": sr}, return_timestamps=False)
+        question = result.get("text", "")
+
+        logging.debug(f"Transcribed question: {question}")
+
+        if not question:
+            return "No transcription available, please try again."
+
+        # Get response using the transcribed question
+        response = get_response(question)
+        logging.debug(f"Response from Neo4j and GPT: {response}")
+
+        # Generate audio from the response
+        audio_path = generate_audio_elevenlabs(response)
+        logging.debug(f"Generated audio path: {audio_path}")
+
+        return audio_path
+    except Exception as e:
+        logging.error(f"Error in handle_voice_to_voice: {e}")
+        return "Error processing the audio, please try again."
+
 
 
 # Define the Gradio interface