Update app.py

app.py

@@ -211,7 +211,7 @@ def generate_music(genre_txt, lyrics_txt, max_new_tokens=5, run_n_segments=2, us
     decoded_instrumentals = codec_model.decode(torch.as_tensor(instrumentals.astype(np.int16), dtype=torch.long).unsqueeze(0).permute(1, 0, 2).to(device)).cpu().squeeze(0)
 
     mixed_audio = (decoded_vocals + decoded_instrumentals) / 2
-    return (16000, mixed_audio.numpy())
+    return (16000, mixed_audio.tensor.detach().numpy())
 
 @spaces.GPU(duration=120)
 def infer(genre_txt_content, lyrics_txt_content, num_segments=2, max_new_tokens=10):