#settings for vonovox
1 messages · Page 1 of 1 (latest)
for male voices keep at 0 if ur a guy
for female voices choose between 3-10 maybe 12 if you're a guy
for pitch
if you're a girl I dunno
are there like other optimizations i can make with like crossfade or index rate or stuff like that?
Index rate keep at 0, crossfade I personally say 0.15 is good, block size is really dependent on ur pc just mess with it until the delay is good for u, and extra at 3.0 or 2.7
If it makes the output sound worse just go back to default
Leave Smart SINE on
Agreed
Matter of fact, just read this: https://docs.aihub.gg/realtime-voice-changer/local/vonovox/#pros--cons
Last update: November 21, 2025
you can also use chatgpt to help you with fine-tuning EQ I guess
If you're doing male-to-female, you can fiddle with the formant setting, too. I think a range of 0-0.8 is a sweet spot to play around with. But it's something that bears experimenting for sure
no the upscaler is applied on the model inference output so it depends on the model sample rate (32k, 40k, 48k) and the input mic sample rate and fidelity aren't relevant, otherwise the upscaler would be useless.
the upscaler won't be effective for 48k models and may have some more performance load, so you can turn it off if you don't need it to save the performance load or you think anything above 16khz isn't so perceivable to distinguish the audio quality.
also you haven't been asked to mention your GPU spec
tagging @sturdy reef@final warren you guys should have asked it first
so if you have like RTX 3060 or lower, and/or you want to run some game, I'd advise to turn off the AP-BWE 48k Upscaler