IndexTTS2: Emotional duration-controlled autoregressive zero-shot text-to-speech

[satvikpendem]

https://index-tts.github.io/index-tts2.github.io/

[satvikpendem]

ArXiv: https://arxiv.org/abs/2506.21619

The demos are very impressive, and the code looks to be Apache 2 licensed just as IndexTTS and IndexTTS1.5 were.