New AI can simulate your voice from just 3 seconds of audio

My interest in not publicly available models is very limited.