Update app.py

app.py

@@ -223,7 +223,7 @@ def detect_voice_activity(audio_file, threshold=0.02):
     if len(data.shape) > 1:
         data = np.mean(data, axis=1)
     
-    # Normalize the audio data
+    # Normalize the audio data to the range [-1, 1]
     if data.dtype != np.float32:
         data = data.astype(np.float32) / np.iinfo(data.dtype).max
     
@@ -248,9 +248,12 @@ def detect_voice_activity(audio_file, threshold=0.02):
     for segment in voice_segments:
         trimmed_audio = np.concatenate((trimmed_audio, data[segment[0]:segment[1]]))
     
+    # Convert the trimmed audio back to 16-bit integer format
+    trimmed_audio_int16 = np.int16(trimmed_audio * 32767)
+    
     # Export the trimmed audio
     output_path = "voice_trimmed_audio.wav"
-    wavfile.write(output_path, sample_rate, trimmed_audio)
+    wavfile.write(output_path, sample_rate, trimmed_audio_int16)
     
     # Clean up the converted WAV file
     os.remove(wav_path)