Update app.py

app.py

@@ -21,7 +21,7 @@ def translate_speech(audio_input):
     input_dict = processor(audio_data, return_tensors="pt", padding=True)
 
     # Use the model to get the logits
-    logits = model(input_dict.input_values.to("cuda")).logits
+    logits = model(input_dict.input_values.to("cpu")).logits
 
     # Get the predicted IDs
     pred_ids = torch.argmax(logits, dim=-1)[0]