Update app.py

app.py

@@ -250,21 +250,21 @@ def predict(
         video = read_video_pyav(container=container, indices=indices)
 
     if uid in generated_audio_files and len(generated_audio_files[uid]) != 0:
-        audio_length_in_s = get_audio_length(generated_audio_files[uid][-1])
+        audio_length_in_s = min(get_audio_length(generated_audio_files[uid][-1]), 30)
         sample_rate = 24000
         waveform, sr = torchaudio.load(generated_audio_files[uid][-1])
         if sample_rate != sr:
             waveform = torchaudio.functional.resample(waveform, orig_freq=sr, new_freq=sample_rate)
         audio = torch.mean(waveform, 0)
-        audio_length_in_s = int(len(audio)//sample_rate)
+        audio_length_in_s = min(int(len(audio)//sample_rate), 30)
         print(f"Audio Length: {audio_length_in_s}")
     else:
         generated_audio_files[uid] = []
     if video_path is not None:
-        audio_length_in_s = get_video_length(video_path)
+        audio_length_in_s = min(get_video_length(video_path), 30)
         print(f"Video Length: {audio_length_in_s}")
     if audio_path is not None:
-        audio_length_in_s = get_audio_length(audio_path)
+        audio_length_in_s = min(get_audio_length(audio_path), 30)
         generated_audio_files[uid].append(audio_path)
         print(f"Audio Length: {audio_length_in_s}")