Microsoft unveils text-to-speech AI model VALL-E, which was trained on English speech data and can simulate a person’s voice with three seconds of sample audio (Benj Edwards/Ars Technica)

Benj Edwards / Ars Technica:
Microsoft unveils text-to-speech AI model VALL-E, which was trained on English speech data and can simulate a person’s voice with three seconds of sample audio  —  Text-to-speech model can preserve speaker’s emotional tone and acoustic environment.  —  On Thursday, Microsoft researchers announced …