Update app.py

app.py

@@ -113,6 +113,12 @@ pipe_asr = pipeline(
 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", "")
 
@@ -123,6 +129,7 @@ def handle_voice_to_voice(audio):
     audio_path = generate_audio_elevenlabs(response)
     return audio_path
 
+
 # Define the Gradio interface
 with gr.Blocks() as demo:
     audio_input = gr.Audio(sources=["microphone"], type='numpy', streaming=False, label="Speak to Ask")