audio_input fix

app.py

@@ -58,6 +58,9 @@ if uploaded_file is not None:
     audio_input = AudioSegment.from_file(temp_wav_path).set_frame_rate(16000).set_channels(1)
     audio_input = np.array(audio_input.get_array_of_samples())
 
+    # Normalize audio
+    audio_input = (audio_input - np.mean(audio_input)) / np.std(audio_input)
+
     # Process the audio
     input_features = processor(audio_input, return_tensors="pt", sampling_rate=16000).input_values