Created by: cogmeta

feature = data_generator.process_utterance(filename, "")

Is there a way to process utterance directly from the buffer without writing first to file and then passing the file as an argument?