torn otter
Yeah you should never trust video tutorials for realtime voice changers
They use an over year old version of original wokada which have worse quality and performance
also, vb audio cable causes issues on windows
You should uninstall and forget every info in youtube tuts
ocean ivy
oh okay is there a tutorial or some special way i should uninstall wokada ? i just uninstalled vbcable through the installer app
im mtf trans and just want to sound more feminine so i dont get slurs thrown at me every time i turn on my mic in a game lol
torn otter
im mtf trans and just want to sound more feminine so i dont get slurs thrown at ...
that's great then, I just asked because some have been caught doing catfishing, which is not allowed and can lead to illegal stuff
ocean ivy
yeah im not surprised
torn otter
oh okay is there a tutorial or some special way i should uninstall wokada ? i ju...
you just delete the old original wokada folder and uninstall the vb audio cable from the windows app settings
-realtime
torn otter
1st link, here you will get wokada deiteris fork and VAC Lite
torn otter
yeah im not surprised
ofcourse we don't help and promote ai for bad purposes lol, you should be shocked how many think it's a "flex" 😭 (ofc they get banned)
ocean ivy
oh wow this is a good step by step ty!! i will follow and let you know if i run into any issues~~
ocean ivy
ofcourse we don't help and promote ai for bad purposes lol, you should be shocke...
they think catfishing is a flex? LMAO
torn otter
alr, lmk
torn otter
they think catfishing is a flex? LMAO
yeah 💀
ocean ivy
wait do i need to restart after installing vac
torn otter
wait do i need to restart after installing vac
You don't have to
ocean ivy
ok i got it! do i just choose any of the models in the voice-models channel
does the higher epoch ones mean its more accurate?
mental nova
does the higher epoch ones mean its more accurate?
nope
uhh it's hard to explain the concept of epochs, but, no, more doesn't always equal better
there are multiple factors really
they put the epoch amount in the title because for some reason, it is mandatory to do so (it is against the server rules not to put the number of epochs), idk why such rule exists lmao
torn otter
does the higher epoch ones mean its more accurate?
epochs are a unit of measuring the training cycles of the AI model
basically the amount of times the model went over its dataset and learned from it
they don't mean how good is the model, it's just an info provided on how they trained the model by the model maker
More ≠ better
Less ≠ better
There's no way to determinate how good the RVC model is until you try it out or listen to the audio samples if there are
ocean ivy
ohhhh okay let me try a few then !
its pretty good for the most part but sometimes its kinda crackly and i cant tell if it's my mic or if its the model
torn otter
its pretty good for the most part but sometimes its kinda crackly and i cant tel...
Show a screenshot of your wokada
ocean ivy
torn otter
Are the cracks only on discord?
mental nova
share an audio so we can hear whats the issue
torn otter
on wokada deiteris fork, you can **optionally **use more advanced settings for benefits:
ocean ivy
Are the cracks only on discord?
oh i havent used it in discord yet i just did the recording from the wokada app itself
ok ill send over a recording in a bit
ocean ivy
ocean ivy
oh okay
ocean ivy
wait sorry dumb question but is it not better if it's trained more?
oh wait that one sounds so much more natural !! how do i download it?
mental nova
wait sorry dumb question but is it not better if it's trained more?
model was probably trained using cringe asmr mommy type data
nothing can be done, the model is fcked up
mental nova
oh wait that one sounds so much more natural !! how do i download it?
i dont share my models, but a rvc model should sound like that
mental nova
how do i know if it's a rvc model or not in the voice models channel?
they have the "rvc" tag
models have an easier life if they're cloning things similar to their voices
male to male conversions = almost perfect
male to female = weird
if they sound like a smoker is because the model is overfitted (having a strong bias to do X sound, in rvc models case) with that particular sound
mental nova
if they sound like a smoker is because the model is overfitted (having a strong ...
and nop, this can't be fixed (well it can be fixed if the model maker retrains the dataset, adding good diverse data and removing trash data)
ocean ivy
ohhhhhh thats so interesting im super new to this world so good to know
is it hard to make your own model?
mental nova
is it hard to make your own model?
nope but to be honest, it can be a bit annoying and time consuming
making models of your favorite game character is for the most part, very easy, because voice lines sound good and are already cleaned by the dev team, so you really dont have to do anything besides clicking the training button
in the other hand, realtime models can be a bit challenging if you just started making models, since you gotta do the audio cleaning process yourself, and for realistic results rvc needs a lot of data (minimum 2 hours)
it also needs to be good data, for example, asmr or soft talk data is horrible for models, but natural speech is perfect
you need a gpu that at least have 8 gb of vram
mental nova
sure! good luck, any questions dont be afraid of asking them in the #✨│ai-help channel, im most of the time there helping new model makers
quartz remnant