requirements

app.py

@@ -244,7 +244,7 @@ def compute_video_embedding(video_reader, min_len, vae_model):
     return encoded, video_frames
 
 
-@spaces.GPU(duration=120)
+@spaces.GPU(duration=60)
 @torch.no_grad()
 def compute_hubert_embedding(raw_audio, hubert_model):
     """Compute embeddings from audio"""
@@ -291,7 +291,7 @@ def compute_hubert_embedding(raw_audio, hubert_model):
     return audio_embeddings
 
 
-@spaces.GPU(duration=120)
+@spaces.GPU(duration=60)
 @torch.no_grad()
 def compute_wavlm_embedding(raw_audio, wavlm_model):
     """Compute embeddings from audio"""
@@ -398,7 +398,7 @@ def extract_video_landmarks(video_frames, landmarks_extractor):
     return np.array(processed_landmarks)
 
 
-@spaces.GPU(duration=600)
+@spaces.GPU(duration=200)
 @torch.no_grad()
 def sample(
     audio_list,
@@ -617,7 +617,7 @@ def sample(
     return complete_video
 
 
-@spaces.GPU(duration=600)
+@spaces.GPU(duration=200)
 @torch.no_grad()
 def process_video(video_input, audio_input, max_num_seconds):
     """Main processing function to generate synchronized video"""