Upload to debug

app.py

@@ -136,7 +136,7 @@ def run_avse_inference(video_path, audio_path):
             est = avse_model.enhance(chunk_data).reshape(-1)
             estimated_chunks.append(est)
 
-    estimated = torch.cat(estimated_chunks).cpu().numpy()
+    estimated = np.concatenate(estimated_chunks, axis=0)
 
     # Save result
     tmp_wav = audio_path.replace(".wav", "_enhanced.wav")