Update app.py

app.py

@@ -106,7 +106,7 @@ codec_model.eval()
 # vocal_decoder.eval()
 # inst_decoder.eval()
 
-
+@spaces.GPU(duration=120)
 def generate_music(
         max_new_tokens=5,
         run_n_segments=2,
@@ -296,13 +296,22 @@ def generate_music(
         vocal_sr = 16000
         instrumental_sr = 16000
         mixed_sr = 16000
+        
+        # added scaling to the audio
+        limit = 0.99
+        max_val = np.max(np.abs(mixed_waveform))
+        mixed_waveform = mixed_waveform * min(limit / max_val, 1)
+
+        max_val = np.max(np.abs(decoded_vocals_waveform))
+        decoded_vocals_waveform = decoded_vocals_waveform * min(limit/ max_val, 1)
+
+        max_val = np.max(np.abs(decoded_instrumentals_waveform))
+        decoded_instrumentals_waveform = decoded_instrumentals_waveform * min(limit/max_val,1)
 
         print("All process Done")
 
         return (mixed_sr, mixed_waveform.numpy()), (vocal_sr, decoded_vocals_waveform.numpy()), (instrumental_sr, decoded_instrumentals_waveform.numpy())
 
-
-@spaces.GPU(duration=120)
 def infer(genre_txt_content, lyrics_txt_content, num_segments=2, max_new_tokens=15):
     # Execute the command
     try: