Update app.py

app.py

@@ -26,8 +26,8 @@ def infer(text_prompt):
     
     with torch.inference_mode():
         speech_output = model.generate(**inputs, do_sample = True, fine_temperature = 0.4, coarse_temperature = 0.8)
-        audio_out = speech_output[0].cpu().numpy()
-
+        #audio_out = speech_output[0].cpu().numpy()
+        audio_out = speech_output[0]
         # Assuming audio_out contains audio data and the sampling rate
         sampling_rate = model.generation_config.sample_rate
         wavfile.write("output.wav", sampling_rate, audio_out)