Replies: 3 comments 2 replies
-
|
out of curiosity what would that mean for you. would the text be available in advance and you only care about generating it 'faster than realtime' or would the words be streaming in one-by-one? |
Beta Was this translation helpful? Give feedback.
-
|
My application doesn't have the text in advance and requires minimal latency, so I need either streaming audio or really, really fast generation. My problem currently is that there is little pretrained TTS systems for Russian. The best one I guess is Silero, but I hate the fact that it doesn't natively support english inserts or any kind of symbols, and just ignores them. |
Beta Was this translation helpful? Give feedback.
-
|
跟我的需求一样，离线转换太慢了，多数的场景是实时或流式的text2speech场景。 |
Beta Was this translation helpful? Give feedback.
-
|
could this support generating voice stream data ? |
Beta Was this translation helpful? Give feedback.
-
This thing is so awesome, but my application is realtime and so not applicable. I was thinking if at some point it could be realtime on high-end hardware (3090 ti in my case) or if it's possible to code and train similar but smaller model on data with one language to get good results. What do you think?
Beta Was this translation helpful? Give feedback.
All reactions