HeyGen's Professional Voice Clone trains on 30 minutes to 2 hours of audio

HeyGen's Professional Voice Clone is trained on 30 minutes to 2 hours of a speaker's recorded audio, according to a post by Rohan Paul on X. Paul reports that the company says the clone preserves breaths, pauses, laughs and accent, and that the result cannot be told apart from the original speaker.

Paul argues that longer recordings capture hesitations and pace changes that short reference clips miss. The claims about output quality are HeyGen's, as relayed in the post; no independent test is cited.