Image: Microsoft #347Microsoft released VALL-E, a text-to-speech model that can mimic a voice with just three seconds of sample data. Source: Impost AI & MLAudio