my mic is crackling in rvc - gui

[little lagoon]

my gpu is 6700 xt, and these are my settings

[barren rose]

open task manager hit details

right click audiodg.exe and set Priority to HIGH

right click it again and press set affinity and only have an EVEN number CPU core set, 2 is what I use. AKA everything unchecked except CPU 2

also

first

Fade length: 0.15

before you do the first

[hearty ledge]

Your inference time is >295ms. Your sample length must be at least 0.30

[little lagoon]

alr

lemme try

[barren rose]

Im pretty sure a 6700xt can run 0.15 sample length easily

at least the last person that tried it with that gpu said it works

[little lagoon]

i mean

[summer torrent]

Ayo? @little lagoon level 2 !!!

[little lagoon]

it works for a bit

but then it starts to crackle

[barren rose]

did you do the fade length max thing

[little lagoon]

i did it

[hearty ledge]

It might be that the new formant shift option is causing performance issues for AMD GPUs

But you can see that inference time is higher than sample length. It doesn't work well if that's the case

[hearty ledge]

Try a bit more sample length probably. 0.35-0.40

[little lagoon]

how can i make it less robot like

[barren rose]

could be the voice model you have is not suitable for realtime

unless youre using the default which should be fine

[little lagoon]

i got it off of voice models

[little lagoon]

do you know hopw to be heard louder?

[hearty ledge]

Increase loudness factor

Or increase volume sound on your mic

[little lagoon]

theres still some sort of cutting off

[barren rose]

if its the beginning then lower Extra

[little lagoon]

no

like

stuttering

like its not consistent

[barren rose]

that means ur gpu is not able to run it which makes no sense

did you run the right file btw lol

[little lagoon]

yea

[barren rose]

is your gpu being recognized

[little lagoon]

how do i check that

[barren rose]

first run go-realtime-gui-dml

[little lagoon]

i ran that

[barren rose]

then after everythihng loaded up take a screenshot of command window

dont hit start conversion yet

[barren rose]

i said

dont hit start conversion lmfao

[little lagoon]

🙂

[barren rose]

i need the top part

[barren rose]

after everything loaded up

💀

[barren rose]

is that all?

for example mine says this

[barren rose]

@hearty ledge anything unusual here?

should we maybe run the command where you can check if the gpu is recognized

(idk what command that was again tho)

[little lagoon]

[hearty ledge]

Doesn't looks like anything unusual, but the logging is different 🤔 It should appear like this

[little lagoon]

im on amd

hmm

[summer torrent]

Ayo? @little lagoon level 3 !!!

[little lagoon]

this the name

of the install

[hearty ledge]

Well, I'm using the same version and I'm also on AMD. Don't have any issues with running it though

[barren rose]

does windows11 not show it or something

[little lagoon]

damn

[little lagoon]

any clues??

or should i use sum else

[barren rose]

you acn try emojikages wokada version

https://github.com/deiteris/voice-changer/releases/tag/b2020

voice-changer-windows-amd64-dml.zip

[little lagoon]

im using rvc

imma try that

[little lagoon]

any guides?

[barren rose]

you only have to run the mmvcserversio.exe file

the rest is the same as the wokada guide except since it has better performance, you can go lower in chunk and higher in extra

[little lagoon]

it sounds weird asf

like nothing like avoice changer

[barren rose]

i dont know how to help with that description

[little lagoon]

like stutters and crackles

@hearty ledge

[hearty ledge]

Could you show settings?

[little lagoon]

and the voice is unusually hard and not soft

[little lagoon]

sure

@hearty ledge

6700 xt

[hearty ledge]

1. You need to change "TUNE" to change the pitch of the voice. It does not adjust automatically.
2. You need to select GPU according to the number you see in Task Manager. If you have only one GPU, select gpu0.

[little lagoon]

rn

thanks it works now

but only with base models wth

are these stats normal?

[little lagoon]

still stutters on models that did not come with the voice changer

thats really weird

[barren rose]

since you have an AMD gpu you need to convert every voice model you upload to ONNX on that grey button next to save, then reupload that newly made onnx file

youre better off using emojikages wokada version that doesnt require this and runs better

[little lagoon]

i tried installing it

can you give me the link again please?

[barren rose]

https://github.com/deiteris/voice-changer/releases/download/b2020/voice-changer-windows-amd64-dml.zip

[little lagoon]

is this this one

yea it is

i was using that

[barren rose]

the name is the same as the normal one

so ignore that

did you DOWNLOAD the same link i sent you is the question

[little lagoon]

yeas

[summer torrent]

Ayo? @little lagoon level 4 !!!

[little lagoon]

lemme just redownlaod ity

[little lagoon]

this is what opened

[hearty ledge]

You should upload a model and select it. It doesn't come with default ones

[little lagoon]

Then i prob got the wrong pne

Then i prob got the wrong one

[barren rose]

no you got the right one

you just simply have to upload a model xd

[little lagoon]

are these settings ight?

@hearty ledge where can i see what settings i should use for your client?

[hearty ledge]

https://rentry.co/RVCRealtimeGuide#gpu-chart-for-sample-length-extra-delay-quality

Chunk is sample length, extra is extra

[little lagoon]

what do you think about the quality?

[hearty ledge]

Sounds fine to me. I'd suggest to keep SilenceFront on in Advanced Settings, unless there are any issues

And crossfade length at 0.1ms

[little lagoon]

thanks! but as a last thing. If its getting choppy should i higher the chunks?

[hearty ledge]

Yes. You should either increase chunk or reduce extra