Update ultravox_processing.py

ultravox_processing.py

@@ -151,7 +151,7 @@ class UltravoxProcessor(transformers.ProcessorMixin):
                 return np.floor((in_len + (2*padding) - (dilation * (kernel - 1)) - 1)/stride + 1)
             def stack_frame_len(T):
                 T_pad = ((T + self.stack_factor - 1) // self.stack_factor) * self.stack_factor
-                return (T_pad + self.stack_factor) // self.stack_factor
+                return ((T_pad + self.stack_factor) // self.stack_factor).astype(int)
             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)