Update ultravox_processing.py

ultravox_processing.py

@@ -152,8 +152,8 @@ class UltravoxProcessor(transformers.ProcessorMixin):
             def stack_frame_len(T):
                 T_pad = ((T + self.stack_factor - 1) // self.stack_factor) * self.stack_factor
                 return int((T_pad + self.stack_factor) // self.stack_factor)
-            nb_encoder_frames = [cnn_out_len(cnn_out_len(feat_len, kernel=3), kernel=3, stride=2) for feat_len in data["audio_len"]]
-            data["audio_token_len"] = [stack_frame_len(nf) for nf in nb_encoder_frames]
+            nb_encoder_frames = cnn_out_len(cnn_out_len(data["audio_len"], kernel=3), kernel=3, stride=2)
+            data["audio_token_len"] = stack_frame_len(nb_encoder_frames)
 
         if text is not None:
             assert isinstance(