Update app.py

app.py

@@ -619,8 +619,10 @@ def infer_example(image_path, audio_path, text, num_steps, raw_image_path, sessi
     if session_id is None:
         session_id = uuid.uuid4().hex
 
+    limited_audio_path = preprocess_audio_first_5s_librosa(audio_path, True, session_id)
+
     image_path, _ = preprocess_img(image_path, image_path, [[720, 400]], session_id)
-    result = infer(image_path, audio_path, text, [[720, 400]], num_steps, session_id, progress)
+    result = infer(image_path, limited_audio_path, text, [[720, 400]], num_steps, session_id, progress)
 
     return result