Currently using realtime voice changer and I am looking to refine a voice model to closer resemble the original voice but it does still come out somewhat skewed because the differences in my voice, even with pitch correction. Any advice at refining/tweeking it to better match the original voice?
#Refining/adjusting voice models
hallow hazel
hidden hinge
Currently using realtime voice changer and I am looking to refine a voice model ...
can you share a screenshot of the entire program settings you're using?
hidden hinge
You want the wokada settings?
yes
hallow hazel
I will be able to provide screenshots in like 4 hours. Like I said I am using wokada but I did see that the guide you posted had vonovox as an option and I do have a 5070ti. How does vonovox pair up against wokada?
hallow hazel
Currently I have formant shift set to 0 but I have been playing with it to see if it comes out better.
hidden hinge
uninstall vb audio cable
set extra to 2.7 if you dont want to risk any cutoff issues on some models (it's a bug present in wokada deiteris fork)
set chunk to 200
on wokada deiteris fork, you can **optionally **use more advanced settings for benefits:
- Advanced Settings -> Force FP32 mode: on (THIS IS OFF BY DEFAULT! Turning this on improves stability. Increases VRAM usage by 200 MB)
- Advanced Settings -> Disable JIT compilation: off for faster loading speed of the program, on for slightly better performance (10-15 ms) for Nvidia only)
- Advanced Settings -> Crossfade Lenght: Controls how smoothly the AI stitches different processed parts "chunks" of your voice back together. 0.1 for fastest voice, 0.15 for improved quality but increases delay by ~50 ms
- Reduce the delay on Windows via the Wasapi / Asio Guide
hallow hazel
Ill try that now, I was playing with vonovox too and trying to play with those setting some and think im dialing into what i like
would you have any suggestions on getting a audio of a laugh to convert right. Even not in real time just a recording if possible or is that purely dependent on the trained model?
hidden hinge
would you have any suggestions on getting a audio of a laugh to convert right. ...
any suggestions on getting a audio of a laugh to convert right
Nope, RVC can't always do super realistic laugh, if the model was trained on some specific laughs it might help, but rvc is limited on non speech sounds
hidden hinge
@hallow hazel soo, is this solved?
hallow hazel
Sorry yes!