Update app.py

app.py

@@ -134,6 +134,7 @@ def extract_speech_ids(speech_tokens_str):
             print(f"Unexpected token: {token_str}")
     return speech_ids
 
+@torch.inference_mode()
 @spaces.GPU(duration=15)
 def infer_with_speaker(
     display_name,
@@ -161,7 +162,7 @@ def infer_with_speaker(
         progress,
     )
 
-
+@torch.inference_mode()
 @spaces.GPU(duration=15)
 def gradio_infer(*args, **kwargs):
     return infer(*args, **kwargs)