I've been thinking for a while about the possibility of having a Yukkuri TTS. I know some Japanese streamers have it where it's used in proper but I was trying to think about it being able to be used in English. The only problem is I'm not sure how to get that connection. It makes sense where it would be a machine learning thing that would see how a word is supposed to sound and then translate that to a similar sounding set of katakana characters for it to read, but I haven't found anything like that thus far. Does anyone know a program that can make it happen or someone who can do this work (can negotiate pay)?
bump
>>1460832https://www.maginative.com/article/openvoice-is-an-impressive-new-open-source-voice-cloning-ai/I just tried the OpenVoice demo a bit for you. First I rendered some English text using AquesTalkPlayer, the original "Yukkuri" voice:https://vocaroo.com/156StZi5TyRxThe OpenVoice demo returned this:https://vocaroo.com/1fHB9w9JUPJ1My guess is that the Yukkuri voice is so low quality that the AI has to make up details, it just doesn't sound the same.I don't have a powerful GPU to try something local. Maybe I should try another demo somewhere else.
>>1461342Hadn't heard of Aques before. In that case, do you think it would be better to have a program that takes chat messages during a redeem, pastes it into Aques, hits plan, and then clears out the text afterward? Sorry I am a bit dumb
