Last year I saw a TED Talk by reps from Adobe that featured a new technology that can synthesize any speaking voice simply by analyzing a twenty-minute recording of the original. It was…freaky.
That’s weird enough, but Lyrebird, a Montreal company, has an API that does the same thing but only requires a one minute recording of the original. This is scary stuff. I’ve clipped this from their website:
COPY THE VOICE OF ANYONE.
Lyrebird will offer an API to copy the voice of anyone. It will need as little as one minute of audio recording of a speaker to compute a unique key defining her/his voice. This key will then allow to generate anything from its corresponding voice. The API will be robust enough to learn from noisy recordings. The following sample illustrates this feature, the samples are not cherry-picked.
Please note that those are artificial voices and they do not convey the opinions of Donald Trump, Barack Obama and Hillary Clinton.
(Click on the graphic to be taken to the example. They won’t let me embed it.)
What could possibly go wrong? No one would ever use this for evil/illegal/immoral purposes, right? More here.