AI can fake anyone's voice in just a few minutes

The founders of Lyrebird, a startup AI (artificial intelligence) in Montreal set a goal of "creating the most realistic artificial voice in the world". And although new at the development stage, their AI has been able to mimic voice with very high accuracy.

Ashlee Vance of Bloomberg interviewed Lyrebird's founders last month. And in this interview, Lyrebird's AI recorded Vance's voice and cloned it within minutes. In order to check the accuracy of the human voice, Vance called his mother and used the voice that the AI ​​duplicated and prepared. Surprisingly, his mother never realized he was talking to the phone, not his son. This shows that Vance's voice created by AI is highly accurate.

Lyrebird's AI is used to help people who lose their voice. Lyrebird uses footage video clips that the person said, previously stated to create a human voice. One of the users of Lyrebird's AI-based human voice is Pat Quinn, founder of Ice Bucket Challenge. Previously, Quinn communicated with a text-to-speech transfer system.

However, many people are worried about the risks and the possibility of being abused by bad guys (criminals or frauds) that the human clone of Lyrebird's exact voice may encounter.

See more:

• Google AI can create 3D images from 'flat' 2D images.
• Nvidia uses AI to create slow motion video 240fps from smooth, lifelike 30fps video
• AI researchers have stopped developing human-like machines