Microsoft trained VALL-E's speech-synthesis capabilities on an audio library, assembled by Meta, called LibriLight. It contains 60,000 hours of English language speech from more than 7,000 ...

Now that people can easily create real-sounding voices with artificial intelligence, detection technologies are racing to catch deepfake audio, but it's a tough game of whack-a-mole.

A chimpanzee skull studied for conspecific recognition. Credit: Kyoto University/André Gonçalves The study was conducted using an experimental booth housed in a testing room.