Update module

app.py

@@ -104,7 +104,6 @@ def run_avse_inference(video_path, audio_path):
 
     with torch.no_grad():
         estimated = avse_model.enhance(data).reshape(-1)
-        print(estimated.shape)
 
     # Save result
     tmp_wav = audio_path.replace(".wav", "_enhanced.wav")
@@ -150,7 +149,7 @@ def extract_faces(video_file):
             # version 2
             h, w, _ = frame.shape
             x1, y1, x2, y2 = box.xyxy[0].cpu().numpy()
-            pad_ratio = 0.3  # 30% padding
+            pad_ratio = 0.5  # 30% padding
 
             dx = (x2 - x1) * pad_ratio
             dy = (y2 - y1) * pad_ratio