This is a project that was made specifically for elpws . He always wanted the AI TTS voices used on streamers like Forsen and xQc. I wanted to help him out, however I didn’t know of any way to use previous AI voices. I didn’t want to train a bunch of voices from scratch, so I didn’t know what to do for a while.
Until I got a whisper from Tajj . I didn’t know he whispered me until a couple days later, but as soon as he told me, I went ahead and got to work.
This project uses the Uberduck.ai API, which basically makes it so that we can use thousands of previously made voices that were trained by the community. There are some pros, and of course, some cons.
Pros:
- Thousands of previously made voices
- Voice training off the PC (no lag for streamer/training is offsite)
- Easy to use for the streamer+chatter
Cons:
- Inconsistent amount of wait times.
- Sometimes the TTS request takes a long time for the request to finish, however sometimes they happen instantly
- Request failures sometimes
- This isn’t that much of a fail, since we are (or have) implemented a way to circumvent this by just trying the request again. (However, again, this increases waiting times)
Like some of my other projects, this one is still in heavy development, so stay tuned to the GitHub Repo and the Discord Server.