So it looks like Adobe just broke existing voice authentication.
Project VoCo, unveiled at the annual IMAX event, takes an audio sample and generates a transcript you can easily edit to make the speaker say whatever you want simply by cutting, pasting, and adding words. As The Next Web notes, it’s “basically Photoshop for audio.”“The algorithm does the rest and makes it sound like the original speaker said those words,” Adobe said in a statement.
You can basically collect audio from someone, create a template of their voice, and then provide an editor where you type what you want them to say. Check out the video.
So imagine just capturing someone’s voice at a cafe and then being able to impersonate them activating their Hey Siri functionality, or gaining access to their work or home that’s secured by voice auth.
And that’s for stuff that could be a problem in the future.
Think about what it does to voice evidence.
First, you’ll be able to create voice evidence which will convince many, many people that you in fact said something.
Second, you’ll be able to deny things where there is voice evidence of you saying it, because you can simply say that it was forged.
You type a phrase, and the computer says it in the voice of the target.