Update app.py

app.py

@@ -12,7 +12,10 @@ def transcribe_audio(audio_bytes):
     # Convert bytes to numpy array
     audio_array = np.frombuffer(audio_bytes, dtype=np.int16)
     
-    input_values = processor(audio_array, return_tensors="pt", sampling_rate=16000).input_values
+    # Cast audio array to double precision and normalize
+    audio_tensor = torch.tensor(audio_array, dtype=torch.float64) / 32768.0
+    
+    input_values = processor(audio_tensor, return_tensors="pt", sampling_rate=16000).input_values
     logits = model(input_values).logits
     predicted_ids = torch.argmax(logits, dim=-1)
     transcription = processor.decode(predicted_ids[0])