so-vits-svc-fork VS hifi-gan

https://github.com/voicepaw/so-vits-svc-fork Seems to be this as someone posted this link in response to a different commenter.

There's no explaination in https://github.com/voicepaw/so-vits-svc-fork README.md sadly.

You can do speech to speech voice translation with the so-vits-fork. Make sure to turn on auto predict and run that through the adobe speech enhancer for a good final result.

This is made with both RVC and SVC models using my own dataset. RVC is the newer technology that generally glitches much less, but it glitches in different places than SVC, so it's still useful to have the SVC model for some phrases.

A quick overview of the main methods people use because they're pretty similar. There's sovits-svc, and RVC (also diff-svc but I don't think it's as widely used). Sovits has more models publicly available but RVC is catching up quick. This is all running off your local Windows PC but there's tutorials in the discord groups in the links above if you need more help or info.

Most people use RVC, so-vits-svc or diff-svc. I prefer RVC, it seems faster.

So-vits - https://github.com/voicepaw/so-vits-svc-fork

It will be interesting to see if the artefacts you noticed persist once we've trained the model for longer and switch to a better vocoder such as HiFi-GAN. (The paper and audio examples use WaveGlow since that's the default of the repository we compared ourselves to.) That said, "choppiness" sounds to me like it might be related to the temporal evolution, in which case it's something that a non-causal, convolutional post-net might be able to smooth over.

Hey, as far as I know this paper is the current SoTA on public data that is open source. Github is here. If you are interested in really getting into speech synthesis, this page has everything (modern stuff on the bottom.)