Descript's Lyrebird is a premium service that "allows you to replace recorded words and phrases with synthesized speech that's tonally blended with the surrounding audio." The interactive samples on the website are amazing — I can't tell the difference between the original voices and the synthetic voices. This could be useful for podcast editing, but also for deepfakes (or maybe not — see update below).
Update 10/16/19 3:21pm PT: A spokesperson for Descript emailed me with some clarifications:
– Lyrebird AI is now a part of Descript — and their voice double product is available in private beta as part of Descript's podcast editing software.
– The feature is called Overdub — and a voice double to be used in overdub can only be made of your own voice, which is important for us to emphasize, as we take potential misuse seriously.
