Adobe Project VoCo: We may soon be able to make people say things they never said. How; Adobe plans to release a new voice-editing software.
Η company unveiled Project VoCo yesterday at its annual MAX event, unveiling a tool which will do for audio what Photoshop does for image editing.
VOCO can be used to introduce new words that the speaker has never said and create completely new, and natural, brilliant phrases.
The technology was presented by researcher Zeyu Jin, who wowed attendees at the MAX event with cutting-edge products still in development. We don't know if VOCO will eventually be released as a product. Currently the Adobe Research team is working with the University of Princeton for the project.
“We've developed a technology called Project VOCO where you can simply type the word or words you want to change or insert into VoiceOver. THE algorithm it will do the rest and make it sound like the original speaker said it,” Adobe told TheVerge.
The new software is designed to help content creators in editing a dialogue or narrative, debugging, or even changing a speech.
Despite its predictable success if it is finally released, from that point onwards it will be very difficult to re-trust the recording of one's speech. On the other hand, it could open up a whole new way of keeping one's voice or using voice in other technologies.
Adobe told TechCrunch that Project VOCO is an example of "voice conversion" rather than speech synthesis.
Jin and Princeton researcher Adam Finkelstein who worked with the Adobe Research team said:
"The goal of the voice converter (VC) is to modify a recording that contains the voice of one speaker, so that it sounds like another speaker, without altering the content of the speech."
Let's say both Google and Microsoft are trying to improve voice conversion technology using other techniques.
Watch the presentation video: