I think the better hack would be to run it into a speech-to-text transcoder, and then use a vocoder to play it back.

You lose the "voice" in the process, but in some situations it might actually be fine