extract the numpy.ndarray from the tuple returned by gr.Audio before passing it to the whisper pipeline and added check in transcribe 6d5604d verified develops20 commited on Jun 9