3 ms·

This seems like it would be useful as a pre-processing step before feeding audio to a speech-to-text engine.