2-3 months ago, I asked HN about whether there were any good open source tools or packages for TTS (Text to Speech) [1]
I went through the answers (thank you) and the one I had most success with was tortoise-tts [2], which was seriously impressive, but tediously slow due to leveraging both an autoregressive decoder and a diffusion decoder afaik.
Given the ever increasing rate of change in the space of generative AI, I feel it's worth re-asking the question: what (ideally open source, but it's not necessarily a deal breaker) TTS tools are you having the most success with?
[1] https://news.ycombinator.com/item?id=34211457 [2] https://github.com/neonbjb/tortoise-tts
https://github.com/coqui-ai/TTS
I can never remember the name but always google: incessant loud chirp of the invasive frog
I installed and tried pico-tts as recommended in that thread IIRC
someone said it was good enough... I don't really think so, for reading long text it gets really annoying and I'm hoping for a bit better