Update app.py

app.py

@@ -73,7 +73,12 @@ def predict_with_model(image, model, processor):
 
 def predict_audio(audio_path):
 
-    waveform, sample_rate = torchaudio.load(audio_path)
+    torchaudio.set_audio_backend("soundfile")
+
+    waveform, sample_rate = torchaudio.load(
+    audio_path,
+    backend="soundfile"
+    )
 
     inputs = voice_processor(
         waveform.squeeze().numpy(),