#voice changer
1 messages · Page 1 of 1 (latest)
What application? There's thousands of AI programs
Share a screenshot and elaborate more your issue, along with your PC GPU and what guide link did you follow
Lemme guess, you followed a YouTube tutorial?
All video tutorials are old asf
You got the old version of the original Wokada, which has worse performance and quality
yes
Also, if you installed vb audio cable, uninstall it
It gives random issues on windows
Also your settings were wrong too
Whatever you learn from that tutorial, forget it
I didn't install
Guides for Programs that use RVC Models in Realtime for Calls/Games
Most suggested. GUIDE
ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE
Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated
Read the 1st guide
The Wokada deiteris fork is newer and got more performance and quality
I DONT KNOW
There are no updated video tutorials
What? You just need to read it
You can delete everything you got off that YouTube tutorial
SO WHAT SHOULD I DO?
Delete everything you got off youtube
Read the 1st link up there, the Wokada deiteris fork
You have to read, there's no updated video tutorial
If you don't read it, you won't be able to get the updated program
I read
Alright, let me know how it goes
SHOULD I DOWNLOAD IT?
No need to scream
Read carefully the guide and download the Nvidia windows version one
Everything you need is in the guide
Don't skip any steps
If you skip steps related to windows Nvidia, you could run into issues ofcourse
Translated? It will be converted in realtime yes
No need to scream all caps
It's fine
?
The first one
ok
You can also use a translator for the guide by the way
The guide is English because that's the main and most used language all over the internet
The great majority of everything on the internet is in english
Which is also the most used language for communication all over the world
You need to delete the version you got
It's okd
Old
Install the version said in the guide
Make your usual devices the default ones instead
Are the devices correct now?
the one you set in the pictures no
you need to set as default your usual devices
I dunno what's your usual devices
So what's the use of this download?
A VAC (Virtual Audio Cable) makes a fake audio device, used to re-route the audio of different programs
In Wokada context, it's used to get the output of wokada as the input in other programs
did you set your usual devices as default one
gpu: ur rtx 4060 laptop
f0: rmvpe without onnx
extra: 2.7
input: microphone
output: line 1
What will the full girl voice be like?
you can set the monitor to headphones to see thatr
but u haven't uploaded any model yet
not every model is super high quality, you need to listen to the samples and test it yourself
help
put rmvpe not rmvpe onnx
check the suggested models in the guide
Select Wolada Deitris Fork. The other two are too old to use.
whats the best F0 Det. to use?
like crepe etc
or a guide on what to use for certain scenarios
What's up? What is your PC GPU? And which W-Okada you're using?
4070ti, 2.0.76
No way, you're using the original version of W-Okada, which is so outdated.
which one is the best?
It's still be W-Okada. There's another better version available. https://rentry.co/forkvoicechangerguide#download-nvidia-on-windows
so the nvidia b2332 one?
Yes.
and what makes this better than w-okada, will it deal with robot/artifact sounds better
Too much questions. But the difference being performance.
i see
ill try
will models sound better or does just help with latency/performance stuff
mostly the model quality issue
Next time, better tell about your GPU and the W-Okada version you're using, because if you just ask for settings without telling me that first of all, I won't be able to identify the settings for you.
no, thats a model problem
the fork its just for perfomance boost
im currently looking for a model that deals with robot/artifact sounds the best
so
Anyways, for anyone who has NVIDIA GPU, the F0 will always be regular rmvpe.

Even if you set chunk and extra settings correctly according to the guide and your perf number, but the audio still sounds robotic, it will always be the voice model itself in the end. Try another better voice model.
While you can set F0 to crepe if you use an old RVC voice model that was trained using crepe and from 2023, it's generally not so preferred.
depends in which f0 the model was trained on
most are trained using rmvpe, so you use rmvpe for them
^nothing else are recommended for voice changer
yes, never use a crepe model for realtime
the perfomance is veery bad
and also crepe can't handle even a slight amount of noise, rmvpe handles noise like a champ
@glad silo