#✨│ai-help
kindred dragon
knotty moth
they just want your money for nothing, otherwise you should consider this free open source alternative
https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork
viral mason
Ew voice.ai, use literally anything else, if u have Nvidia u can use any of these, if u don't have Nvidia use either tg fork or Deiteris fork
The ones here are free and better
-rt
heady grove
any like
real time ai voice changer
where i can upload a voice model
and it voice change in real time
knotty moth
real time ai voice changer
just see above you
tawdry cypress
@heady grove The deiteris-w-okada fork is pretty good. Adjust the settings in the "finding my own settings for chunk" section to make it much smoother sounding.
https://docs.aihub.gg/realtime-voice-changer/local/deiteris-w-okada-fork/#finding-my-own-settings-for-chunk
neat bear
I downloaded the dirteris fork for amd like 6 months ago, were there any big updates since then
leaden island
I downloaded the dirteris fork for amd like 6 months ago, were there any big upd...
Whats that?
neat bear
w okada but its a fork for amd I think
hallow thistle
Try open Task Manager and end this program task. There are better realtime voice changer programs than this one.
hallow thistle
I downloaded the dirteris fork for amd like 6 months ago, were there any big upd...
There are like 4 different realtime voice changer versions as of now. The most recent version of Deiteris fork W-Okada is b2332. Another fork that function identically as Deiteris but only made to run on RTX 50 GPU is b2335. Tg Develops' fork has more features and the most recent one, forked from Deiteris W-Okada. Vonovox is the different; it would give much better quality, but its GUI is less friendly and it's only known to work with NVIDIA.
hallow thistle
where i can upload a voice model
!howtoask
hallow thistle
low shard
Applio Lightning.AI exists too btw
also, the refresh method doesn't work? wdym?
the issue is related to ngrok lowered the free tier requests per min
low shard
Applio Lightning.AI exists too btw
@nocturne mural btw not sure if you're aware of https://docs.aihub.gg/rvc/cloud/applio-lightning-ai/ , but lightning.ai allows free tier web uis + gives good hours monthly
I'm not sure if I should PR maybe later the Lightning.AI notebook, because there's no risk of me having the studio template public and would just make it easier for users to use it rather than upload it manually everytime
also welcome back to see you vidal, it's been a while
knotty moth
the issue is related to ngrok lowered the free tier requests per min
as said above it still works by "waiting for a minute"
low shard
@nocturne mural how much time did it take for your horizon tab to get inactive? I never seen that before either and can't find any info related to it
potent bone
how to use
low shard
how to use
Hello, please check: https://docs.aihub.gg/rvc/local/applio/#training
balmy junco
There are like 4 different realtime voice changer versions as of now. The most r...
"CUDA is not available - voice conversion will not work"
Anyone encountered this error while launching Vonovox?
I have CUDA, what packages are missing?
hallow thistle
"CUDA is not available - voice conversion will not work"
Anyone encountered thi...
Is Vonovox the latest version? What is your PC GPU? And what are you trying to do with the program?
balmy junco
Is Vonovox the latest version? What is your PC GPU? And what are you trying to ...
I just git cloned Vonovox from the code itself. What's the way to confirm if it's the latest version?
I'm only launching Vonovox and it showed that error.
hallow thistle
I just git cloned Vonovox from the code itself. What's the way to confirm if it'...
You can try download the zip directly from GitHub. https://github.com/dr87/Vonovox/releases
balmy junco
You can try download the zip directly from GitHub. https://github.com/dr87/Vonov...
I think that's the same as git cloning from this:
I have installed and setup everything
low shard
I just git cloned Vonovox from the code itself. What's the way to confirm if it'...
be sure your gpu drivers are up to date via the nvidia app
and that your windows is up-to-date too
if you still get issues, it might be a random issue that you should fix via the precompiled installation: https://docs.aihub.gg/realtime-voice-changer/local/vonovox/#precompiled-setup-nvidia-on-windows
balmy junco
if you still get issues, it might be a random issue that you should fix via the ...
oh the link got taken down
Precompiled Version of Vonovox
low shard
oh the link got taken down
Precompiled Version of Vonovox
it was a typo in the docs, I'll fix it rn
here's the current latest version: https://huggingface.co/dr87/vonovox/blob/main/Vonovox169.zip
hallow thistle
Remove the /latest in the url and there's this.
balmy junco
thank you both
hallow thistle
You're welcome.
low shard
it was a typo in the docs, I'll fix it rn
here's the current latest version: ht...
The type has been fixed in the docs
low shard
thank you both
yw, lmk
balmy junco
I wish we could post screenshots here
low shard
All good now.
do you need any other help?
viral mason
also, the refresh method doesn't work? wdym?
I tried multiple times and either I'm not doing it right or it just doesn't work consistently
viral mason
Applio Lightning.AI exists too btw
I would need help relearning how to use lightning AI but I'd assume it would have the same issues unless it doesn't require ngrok
low shard
I would need help relearning how to use lightning AI but I'd assume it would hav...
unless it doesn't require ngrok
there are multiple tunnels, like 5, and you can use the normal gradio tunnel instead
low shard
I tried multiple times and either I'm not doing it right or it just doesn't work...
maybe you didn't wait exactly 1 minute
fossil sage
how can i make this sound better if u respond pls ping me
buoyant flame
Best Voice Changer for my specs?
Lenovo Legion 7i
RTX 4070
i9-14900HX
Thanks in advance
viral mason
> unless it doesn't require ngrok
there are multiple tunnels, like 5, and you ca...
I don't know of any tunnels as I didn't even know Ngrok existed until I had to move to kaggle
viral mason
Best Voice Changer for my specs?
Lenovo Legion 7i
RTX 4070
i9-14900HX
Nvidia, AMD, or Intel
buoyant flame
Nvidia, AMD, or Intel
as in GPU or CPU?
low shard
Best Voice Changer for my specs?
Lenovo Legion 7i
RTX 4070
i9-14900HX
RVC isn't the same as a realtime voice changer for calls
RVC = Retrieval-based-Voice-Conversion, the best Few Shots Speech To Speech AI Models (on v2), Inferences (use models) pre-recorded audio (ai covers) and train (make) models. Technically, Mainline RVC does have a go-realtime.bat (aka RVC-GUI), but it's pretty messy and outdated so it's extremely not suggested for realtime. There also updated forks with extra features like Applio.
Wokada = uses RVC for realtime inference. There's 2 main versions, Original made by Wok, and the most suggested one is Deiteris Fork (modified version)
Vonovox = Another Realtime Voice Changer based on RVC, with similar quality and performance to wokada deiteris fork but other perks
low shard
I don't know of any tunnels as I didn't even know Ngrok existed until I had to m...
The lightning.ai guides explains each of them, give it a try, gradio is the default one that's used even in google colab
buoyant flame
RVC isn't the same as a realtime voice changer for calls
RVC = Retrieval-based-...
Ahh, thank you for clarifying, forget the RVC part
low shard
Ahh, thank you for clarifying, forget the RVC part
so, just realtime voice changer for calls/games?
buoyant flame
Yeah
low shard
Vonovox is suggested for your setup
low shard
Nvidia, AMD, or Intel
He said he has an RTX 4070, which is an Nvidia GPU
buoyant flame
Vonovox is suggested for your setup
I was going to go for that, but I heard PC and Laptop GPUs are different so I wanted to see if I should actually go with that based on my specs
viral mason
I don't have the greatest memory but I'll try keeping that in mind for next time
low shard
I was going to go for that, but I heard PC and Laptop GPUs are different so I wa...
A Laptop is a PC (Personal Computer), you're most likely confusing the PC term with the Destkop type
It is true that laptop gpus are weaker compared to their desktop counterpart, but you still got a good GPU for this program, just not as good as an rtx 4070 desktop version
buoyant flame
A Laptop is a PC (Personal Computer), you're most likely confusing the PC term w...
Learnt something new, I'll keep that in mind. Thank you
viral mason
The lightning.ai guides explains each of them, give it a try, gradio is the defa...
When I get home today I'll ping u if that's alright, could u send the guide then for me to read over?
low shard
Learnt something new, I'll keep that in mind. Thank you
You're welcome, let me know :)
buoyant flame
You're welcome, let me know :)
Will do!
leaden island
Anyone here from South East Asia Countries? I would love to connect with you folks! 😃
leaden island
Anyone here from South East Asia Countries? I would love to connect with you fol...
Should i put this text here? 🤔
nocturne mural
<@989772388508000306> how much time did it take for your horizon tab to get inac...
That's what I'm trying to find out, but an estimate would be about 30 minutes.
low shard
That's what I'm trying to find out, but an estimate would be about 30 minutes.
was it 30 minutes of absolute nothing, or were you training/inferencing in the background?
nocturne mural
was it 30 minutes of absolute nothing, or were you training/inferencing in the b...
It was probably the second one; in TensorBoard I had the auto-refresh option enabled, which I thought would keep the page 'active', but still the URL just died.
Or maybe the limit isn’t about inactivity, but rather about how long that URL can stay active.
viral mason
Anyone here from South East Asia Countries? I would love to connect with you fol...
This should probably be in general chat ^^
quartz fog
Heya, wanted to ask what the current best pogramm is for Voice changer with using AI Models
low shard
It was probably the second one; in TensorBoard I had the auto-refresh option ena...
That's weird, should I maybe research other free tunnels?
nocturne mural
That's weird, should I maybe research other free tunnels?
zrok
low shard
I didn't have that issue with horizon personally, tried myself
nocturne mural
low shard
zrok
I'm using Awesome Tunnelling for finding maybe better alternatives, i will see
@nocturne mural wait Gradio Public URL doesn't crash Kaggle anymore? they even embed it (this is a test notebook)
lemme check with applio rq
nocturne mural
I didn't have that issue with horizon personally, tried myself
Have you tried using it for training? Of course, for some cases you can just do a few interactions and that's it, but if, for example, I want to enable the custom pretrained option, or during inference I want to enable some extra process, the interface just crashes.
nocturne mural
Have you tried using it for training? Of course, for some cases you can just do ...
And by crashing, I mean that the interface simply doesn’t update properly, and it doesn’t throw a connection error, which makes me think that sometimes some Gradio requests are just ignored.
random ravine
Are there models that can somewhat clone African or Spanish singing?
low shard
<@989772388508000306> wait Gradio Public URL doesn't crash Kaggle anymore? they ...
@simple ore @nocturne mural Good news! Gradio fixed the Kaggle crash bug:
https://github.com/gradio-app/gradio/issues/6132#issuecomment-2842251870
nocturne mural
<@155030383648440320> <@989772388508000306> Good news! Gradio fixed the Kaggle c...
low shard
Should I just add Gradio as the default value? or do I need to add any other tunnels?
I feel like gradio is the best
nocturne mural
But it would still be necessary to create tunnels for the filebrowser and tensorboard.
viral mason
What is this 💔
viral mason
Is it another potential fix for the Kaggle stuff?
nocturne mural
<@155030383648440320> <@989772388508000306> Good news! Gradio fixed the Kaggle c...
.
nocturne mural
But it would still be necessary to create tunnels for the filebrowser and tensor...
.
viral mason
I'm gonna wait for the new link before I try anything, Kaggle link that is
For applio
nocturne mural
well lemme see with zrok
I still think it would be fine to leave ngrok for the filebrowser and tensorboard, although in both cases the limit error shows up later. If you don’t enable auto-update in tensorboard, it can last for quite a while.
"Zrok is a bit more complex to use, I didn’t really understand the docs very well, but as far as I know, you have a limit on creating multiple tunnels, and it just blocks you. However, their system has a way to reuse a tunnel you had already stopped using, which requires you to run a command to recover that tunnel and blah blah blah.
Although for me, it would be enough to just use wandb instead of TensorBoard, and for the filebrowser I would use Horizon Tunnel, LocalTunnel, or InstaTunnel.
low shard
I still think it would be fine to leave ngrok for the filebrowser and tensorboar...
is filebrowser really needed?
users can upload most things from applio itsself, like the audios and models
can't users upload their dataset with applio's dataset maker option, and then download the checkpoints and index via the kaggle output?
nocturne mural
is filebrowser really needed?
users can upload most things from applio itsself, ...
I see it as necessary in order to make minimal modifications, for example, for someone who wants to adjust configurations, experiment with the code, or maybe speed things up, because Kaggle’s filebrowser is somewhat slow, and on top of that, Gradio isn’t exactly that reliable when it comes to uploading things.
For uploading things to Kaggle, I just upload my files to Google Drive or some file hosting site, and then download them through Kaggle.
Although you can upload files through the filebrowser, with the free ngrok plan it’s simply not recommended. So, the filebrowser might be useful for some and not for others, but I would leave it there.
leaden island
Anyone can help me, Twilio doesn't have my country code number (including. Singapore +65 ) phone contact no.
it is for my AI Automation Social Media where I create Whatsapp, IG & FB Messenger (Convocore.ai [Chatbot] + Twilio)
dire crystal
Good morning, I'm Brazilian and I'm using Google Translate. Is there a way to make voice calls with an AI?
leaden island
Good morning, I'm Brazilian and I'm using Google Translate. Is there a way to ma...
its a no code workflow
dire crystal
You can try use bland.ai / vapi.ai
Trying to use it to create an AI for this?
leaden island
Im still learning to create & deploy one
tawny radish
is forked wokada better or vonovox?
ember yew
Hello, I just joined in since I'm looking for a program that does what I'm looking for.
I just trained my first voice model (Voice A) using Applio. I have a bunch of pre-recorded audio using a different voice (Voice B). I am looking for a program that has Voice A mimic the Voice B audio, since I assume it would sound more life-like than using text-to-speech.
I could try using a virtual microphone to feed back the Voice B lines as input, but I was just wondering if there is a cleaner method to do this.
ember yew
I think a better way to word this question now is "What are some self-hosted alternatives to ElevenLabs's Voice Changer feature?" as it "allows you to convert one voice (source voice) into another (cloned voice) while preserving the tone and delivery of the original voice."
analog obsidian
I think a better way to word this question now is "What are some self-hosted alt...
go to applio's inference tab, select your model, upload the clip you want to convert
ember yew
I don't know why I thought it would be more complicated than that, but thank you
viral mason
Hello, I just joined in since I'm looking for a program that does what I'm looki...
You're looking for what's called Speech to Speech
That's the simplest way to describe what you were asking for, I see u got help tho so I'm gonna disappear
viral mason
@nocturne mural you think you would be able to port applio over into lightning.ai? No rush of course
It seems it's a better option than Kaggle
No encryption needed, Lyery said that at least
nocturne mural
https://discord.com/channels/1159260121998827560/1159290096458149938/14191270162...
"hidden channel"
nocturne mural
lemme screenshot the messages
Likewise, he already mentioned it to me before."
nocturne mural
do you have plans on adding it at some point?
Actually, I used it a while ago, and now that I’ve tried it again, I just got a PortAudio error, which is normal because of the realtime part. But when I tried to fix it, I ran into another error, so I just had to manually edit the code and remove the realtime from app.py.
viral mason
so realtime causes bugs in the code?
nocturne mural
so realtime causes bugs in the code?
just by removing it the interface would start correctly.
viral mason
is that specifically just for lightning ai or does it work universally with all sites
nocturne mural
is that specifically just for lightning ai or does it work universally with all ...
I wouldn’t know the correct answer, but likewise, I’ll try again in a while and then let you know
viral mason
good luck! and thank you for being so helpful and nice
nocturne mural
good luck! and thank you for being so helpful and nice
nocturne mural
good luck! and thank you for being so helpful and nice
When creating a project, you must configure the Python environment by clicking the first button on the right sidebar, 'Environment'. Then click on the Python version (3.10). A dropdown will appear; select 3.11.13 and that’s it.
viral mason
I tried using it on kaggle and it didn't work, is it only for lightning?
nocturne mural
I tried using it on kaggle and it didn't work, is it only for lightning?
Well yes, the Kaggle path is /kaggle/working/, not /teamspace/studios/this_studio/
xd
viral mason
Well yes, the Kaggle path is /kaggle/working/, not /teamspace/studios/this_studi...
I'm a little slow ❤️
nocturne mural
I'm a little slow ❤️
there’s no point in running it on Kaggle since TensorBoard doesn’t work there; that’s why tunnels are used.
viral mason
there’s no point in running it on Kaggle since TensorBoard doesn’t work there; t...
ohh
viral mason
When creating a project, you must configure the Python environment by clicking t...
I'll be trying this in the morning
fallen yoke
Hi guys, can i ask a question? i already know how to train a model and everything, i just wanna ask what happens if i took a dataset that has main vocals and adlibs in the same audio. It can still turn out good or it has to be only main vocals??
viral mason
Hi guys, can i ask a question? i already know how to train a model and everythin...
What does this mean
dawn pulsar
why does my voice changer sounds weird
low shard
Although you can upload files through the filebrowser, with the free ngrok plan ...
Alright
low shard
is forked wokada better or vonovox?
what's your pc gpu and operating system? also be aware that fork just means modified version in tech field
low shard
<@989772388508000306> you think you would be able to port applio over into light...
I already did that
I sent you the link before
low shard
https://docs.aihub.gg/rvc/cloud/applio-lightning-ai/
@nocturne mural are you aware of the lightning.ai notebook?
leaden island
Anyone can help me, Twilio doesn't have my country code number (including. Singa...
.
viscid topaz
no e girl trolling is allowed ⚔️
balmy junco
@low shard What other ways can be used to match accent ? Index file and tuning the index meter hardly affects it
viscid topaz
To maintain a legal, safe & ethical community, we will NOT provide help for:
(E girl, as an example) catfishing/trolling, scamming, impersonation.
Any illegal activities.
Requests for these topics will be ignored and may result in moderation action.
tired escarp
**To maintain a legal, safe & ethical community, we will NOT provide help for:**...
alr mb u right
viscid topaz
yep thanks for understanding!
tired escarp
is there a app thats made for amd gpus or are most of them made for nvidia gpus
dawn pulsar
is there a app thats made for amd gpus or are most of them made for nvidia gpus
i think theres a amd version of voice changers
ruby idol
what voice changer supports .pth files?
viral mason
what voice changer supports .pth files?
all of them 
what gpu do u have, Nvidia, AMD,or Intel
ruby idol
what gpu do u have, Nvidia, AMD,or Intel
nvm i found one i just need to get better ping
viral mason
nvm i found one i just need to get better ping
u sure? the only up to date good ones are here all I need to know is your gpu
-rt
viral mason
Vonovox is Nvidia only but Wokada Tg and Wokada Deiteris work for all gpus
ruby idol
u sure? the only up to date good ones are here all I need to know is your gpu
its nvidia
i just need better ping i think
viral mason
did u get the one you're using from a yt video
if u did it's over a year old maybe 2
outdated and basically really bad
any of these three would work for u and I would suggest since u use nvidia to get vonovox
ruby idol
did u get the one you're using from a yt video
yeah
viral mason
it's the best out thatas of rn
viral mason
u download it, extract, run the setup.bat then after that finishes run the start.bat
not too complicated
plus if u get lost or anything I'm here to help
ruby idol
is the official download link like github dr87 vonovox releases
oh nvm
is it a huggingface link
viral mason
is the official download link like github dr87 vonovox releases
should be here https://github.com/dr87/Vonovox/archive/refs/tags/v1.6.9.zip
most recent one
ruby idol
i got that one
now what input and output device do i put on the voice changer, but on discord?
viral mason
now what input and output device do i put on the voice changer, but on discord?
did u download vac lite? it was also in the guide
ruby idol
did u download vac lite? it was also in the guide
didnt, ill download it now
viral mason
I use a diff setup than normal users of the voice changer so I wouldn't be for sure on discord but for Vonovox
Input: regular headset/headphone mic
Output: Line 1 (vac lite)
ruby idol
can i use my mic that i have instead of my headphones one?
viral mason
can i use my mic that i have instead of my headphones one?
any mic man, just not a voicemod mic as that breaks it
ruby idol
ok
ruby idol
any mic man, just not a voicemod mic as that breaks it
btw where do i download vac lite from
viral mason
btw where do i download vac lite from
just click on the deiteris guide should be at the very beginning
idk if it's possibly missing from the Vonovox guide or not
ruby idol
VAC lite (virtual-audio-cable by muzychenko) right?
viral mason
yup!
ruby idol
thanks for the help
viral mason
thanks for the help
np! if u need any more help just lemme know
native compass
Can someone please help me
when im using the voice changer the default voices thats built into the app works
but when i try a model from #1175430844685484042 it just dont work
@viral mason maybe you can help?
please?
viral mason
the default voices huh
yea you got an old outdated youtube tutorial voice changer didn't you
viral mason
<@1023278814752677918> maybe you can help?
what's your gpu, Nvidia, AMD, or Intel
native compass
what's your gpu, Nvidia, AMD, or Intel
nvidia but an old one 😭
native compass
yea you got an old outdated youtube tutorial voice changer didn't you
idk 😂
viral mason
nvidia but an old one 😭
do u know the numbers orr
I mean u can check here in task manager
viral mason
so u have 3 options u can download but I'd recommend Vonovox first as it's the best of the 3
ruby idol
never hear of a quadro
hey i tried to start the voice changer but its stuck on "warming up voice conversion please wait"
viral mason
the guides are all here and u only need two things, vac lite and vonovox
viral mason
hey i tried to start the voice changer but its stuck on "warming up voice conver...
how long has it been doing that?
ruby idol
how long has it been doing that?
1 min and a half
native compass
so u have 3 options u can download but I'd recommend Vonovox first as it's the b...
i already downloaded everything and I got the custom voice model in and all its just when i hit start it wont work like i dont hear anything
but when i use the models thats already in the app it works perfectly fine
viral mason
wdym u already downloaded it
ruby idol
soooo
viral mason
they never read the rules!!!
ruby idol
how do fix this
viral mason
1 min and a half
just try restarting the program
ruby idol
close the cmd?
viral mason
I had that issue before and it fixed itself at some point
viral mason
it did?
viral mason
that's normal
ruby idol
and how do i fix the voice breaking?
ruby idol
like the voice breaks
ruby idol
what voice are you using
postal dude
viral mason
thank god
viral mason
postal dude
could u send the link of which one u chose from the voice models section in the server
viral mason
but it is delayed
0.03 block size, 2.0 second extra, 0.15 crossfade try these and if the block size doesn't sound good for u swith it to 0.30
try this one, that one is also old 😭https://discord.com/channels/1159260121998827560/1381824129854079069
ruby idol
try this one, that one is also old 😭https://discord.com/channels/11592601219988...
😭 why do i always pick the old ones
viral mason
use the search bar and look for the most recently made versions of any voice
viral mason
😭 why do i always pick the old ones
just unlucky ig
anything that uses any pretrain that isn't og or Legacy core is really outdated and has issues
tawny radish
what's your pc gpu and operating system? also be aware that fork just means modi...
rtx 4080 super, windows 11 pro
ruby idol
just unlucky ig
one more thing i cant hear anything with the output device being put on the vac lite
viral mason
one more thing i cant hear anything with the output device being put on the vac ...
i'd just switch to vb cable then since u most likely already have that
it's not recommended to use it at all since it causes issues on windows but it works properly unlike vac lite
low shard
rtx 4080 super, windows 11 pro
vonovox would be more suggested for your setup
ruby idol
i'd just switch to vb cable then since u most likely already have that
and another problem i have the vb output cable at input devices and the vb input cable at output devices
tawny radish
when i use it, it has even more delay
viral mason
and another problem i have the vb output cable at input devices and the vb inpu...
the uh what now
ruby idol
yeah
viral mason
when i use it, it has even more delay
0.03 block size, 2.0 second extra, 0.15 crossfade try these and if the block size doesn't sound good for u swith it to 0.30
try those
tawny radish
0.03 block size, 2.0 second extra, 0.15 crossfade try these and if the block siz...
damn ill try rn thanks!
im glad some people are still actually trying to improve this voicechanger
viral mason
np! if they make delay worse or quality changes ss your original settings and revert back
tawny radish
i was confused when the owner of vonnvox said to keep it the same
oh
my
god
its like crackling really bad 😭
oh
my
god
i fixed it
holy shit
ruby idol
how
viral mason
its like crackling really bad 😭
the crackling of doom and suffering
tawny radish
i fixed the crackling
WHO DIDNT TELL ME ABOUT VONOVOX BRO 😭
this is the method
i was gonna pay someone 600$ to reoptimize w-okadas code and make it have less delay but ig i dont need to now
bc ik the codes a genuine mess
viral mason
i was gonna pay someone 600$ to reoptimize w-okadas code and make it have less ...
bro..
you just got saved
tawny radish
bro..
u dont understand my dedication with w-okada 😭
ive been on this community for years
pretty sure i was here when w-okada released
viral mason
that would have been a crazy loss of money lmao
tawny radish
ehhh not rly
knotty moth
i was gonna pay someone 600$ to reoptimize w-okadas code and make it have less ...
any kind of paid commission or other stuffs are not allowed
tawny radish
any kind of paid commission or other stuffs are not allowed
not from here, i was paying someone through fivver
😭
i got told i could do it by somebody
a diffrent mod im prettysure or contributor
ima lowkey pay for vonovox this is genuinley good
knotty moth
not from here, i was paying someone through fivver
then it's your own responsibility 
hallow thistle
Because I don't have a good PC with decent GPU, so I can't say if either Vonovox or Deiteris/Tg Develops W-Okada has better audio quality, aside from seeing some here saying "Vonovox is better" even if the GUI is more of professional and quite less familar than any other W-Okada.
tawny radish
then it's your own responsibility <:GoldshipShrug:848322940197928980>
yeah ik 💔
tawny radish
Because I don't have a good PC with decent GPU, so I can't say if either Vonovox...
honestly vonovox has better noise suppression
dont get me wrong like i have a loud ass keyboard and its not picking it up like w-okada did
viral mason
Because I don't have a good PC with decent GPU, so I can't say if either Vonovox...
tbh it's like a mind trick I think but Vonovox makes all models sound more natural and realistic
you can roll your r's and stuff
tawny radish
tbh it's like a mind trick I think but Vonovox makes all models sound more natur...
yes bro this is what i needed
knotty moth
ima lowkey pay for vonovox this is genuinley good
the paid features are just post-processing effects
viral mason
so much more natural sounding
tawny radish
ima buy his patreon
viral mason
ima buy his patreon
WAIT NO
tawny radish
the paid features are just post-processing effects
ehh i still wanan support him
viral mason
I mean if u want to
viral mason
why can i get it for free?
u can get around it by downloading fl studio, getting vb cable and voice meeter and voicemod, then connect them all, it causes slightly more delay but u can get better fx for free this way
hallow thistle
If someone here be able to benchmark, record audios from both Vonovox and Deiteris W-Okada and then compare them, I'd appreciate it. 
knotty moth
ehh i still wanan support him
they usually have patreon, kofi, or something I suppose
analog obsidian
is possible to get realtime vst but it takes a bit of time to do the setup
(its actually easy tho)
tawny radish
u can get around it by downloading fl studio, getting vb cable and voice meeter ...
well id still wanna support him for what he does bc i know how dead ai hub or w-okada its self is
nobody has worked on improving it for a while
analog obsidian
hes a cool guy tbh he deserves the money
viral mason
I'm not stopping from supporting him but just saying that u could get better fx that way, at a small cost of teeny tiny extra delay
tawny radish
I'm not stopping from supporting him but just saying that u could get better fx ...
id want to have a more realistic effect with the voicehcanger if ykwim
like you know how voicemeeter works and stuff? id want it so it can sound more realisic by making it seem like your mic is compressed a bit
if u get what im saying
analog obsidian
I'm not stopping from supporting him but just saying that u could get better fx ...
maybe they just wanna support the dev
dr always keeps an eye in vonovox, when a issue is found he fixes it extremely quick
fossil sage
how long should an rvc model be when you train it
tawny radish
how long should an rvc model be when you train it
like sample wise?
fossil sage
wow you responded in one second
tawny radish
u can do less but
it usually sounds pretty good if you train it right, just get good samples
20mins is even good
viral mason
like you know how voicemeeter works and stuff? id want it so it can sound more r...
yea! all u need is voice meeter, vb cable, vac lite, fl studio (i got a free one for u) and voice mod, then u can download whatever fx u want
analog obsidian
i usually use batch size 8, a good dataset and train around 40 epochs
tawny radish
yea! all u need is voice meeter, vb cable, vac lite, fl studio (i got a free one...
dw i cracked fl studio
im never paying for that
viral mason
I wish I was fl studio/j
tawny radish
😭
fossil sage
dw i cracked fl studio
why do you need fl studio for
hallow thistle
The paid Virtual Audio Cable offers you to have 256 VACs at once.
tawny radish
The paid Virtual Audio Cable offers you to have 256 VACs at once.
wtf
analog obsidian
why cracking fl studio when they have a free version? lol
tawny radish
who needs that??
fossil sage
20mins is even good
ok thanks also why do some people say you need 6 hours
viral mason
why cracking fl studio when they have a free version? lol
the free one sucks ass
tawny radish
why cracking fl studio when they have a free version? lol
i feel more tuff doing it
tawny radish
ok thanks also why do some people say you need 6 hours
you dont..
who said that wtf
im pretty sure
analog obsidian
the free one sucks ass
its the same shiet you just cant load projects
tawny radish
thats how long u train it
fossil sage
you dont..
@teal ferry
ok bruh sorry for pinging u 2 times btw let me know if its annoying
tawny radish
it took me 6-7 hours to train a model
viral mason
its the same shiet you just cant load projects
which is bad if u wanna use a very specific setup with fx you finetuned to sound good
tawny radish
ok bruh sorry for pinging u 2 times btw let me know if its annoying
nah its not dw
analog obsidian
which is bad if u wanna use a very specific setup with fx you finetuned to sound...
uh fair
fossil sage
nah its not dw
im talking about the fact that i pinged eleven 2 times
tawny radish
im talking about the fact that i pinged eleven 2 times
ohhhh
analog obsidian
big datasets are better it's true tho
tawny radish
bigger datasets are good sometimes
viral mason
the largest dataset I ever worked with is like an hour
fossil sage
what you think of this video
knotty moth
ok thanks also why do some people say you need 6 hours
u mean the training time not the dataset length?
fossil sage
u mean the training time not the dataset length?
i mean the length of audio file that you are training
tawny radish
yes]
analog obsidian
they need to be diverse enough, having enough pitch variations and words/sentences that do not get repeated often
tawny radish
whats your gpu
4080 super
fossil sage
tuff gpu
hallow thistle
ok thanks also why do some people say you need 6 hours
This is more like how long it takes to finish training a voice model, which depends on the batch setting and how fast your GPU is, not really the overall length of dataset audio.
tawny radish
planning to get a double 5090 for voicemodel training + using the voicechanger for content and stuff
without lag
fossil sage
if you had a 970 it would be immpossible to do realtime voice changer
tawny radish
bc when i use w-okada with valorant, roblox etc it lags soemtimes
tawny radish
if you had a 970 it would be immpossible to do realtime voice changer
lollll i bought this gpu mostly for w-okada
torpid prairie
what if i have an intel gpu which one do i download?
fossil sage
This is more like how long it takes to finish training a voice model, which depe...
oh i see
analog obsidian
the directml version
viral mason
they need to be diverse enough, having enough pitch variations and words/sentenc...
with a dataset where the voice is kinda robotic (on purpose) would it still sound good with a lot of different pitch differences like GLaDOS
fossil sage
This is more like how long it takes to finish training a voice model, which depe...
do you use vocal remover of ultimate vocal remover
analog obsidian
oh wait nvm i forgot intel gpu doesn't work with the voice changer
torpid prairie
oh wait nvm i forgot intel gpu doesn't work with the voice changer
it doesnt?
dang
hallow thistle
do you use vocal remover of ultimate vocal remover
UVR5
knotty moth
i mean the length of audio file that you are training
umm the longest I've ever had is almost 3 hours but also I think it doesnt have noticable difference with the 1 hour one, except the other factors like quality consistency
viral mason
I made a model of her not too long ago and she sounded fine
analog obsidian
nope yea
viral mason
yo dude
just use thishttps://colab.research.google.com/github/Eddycrack864/UVR5-NO-UI/blob/main/UVR5_NO_UI.ipynb?authuser=1#scrollTo=gmjUWmz8iecd
fossil sage
umm the longest I've ever had is almost 3 hours but also I think it doesnt have ...
ok moral of the story make it 1 hour
viral mason
it's uvr5 but not local
fossil sage
just use thishttps://colab.research.google.com/github/Eddycrack864/UVR5-NO-UI/bl...
YOOOOOOOOOOOOOOOOOOOOOO
analog obsidian
i'd recommend vast ai coz is way cheaper, 0,150$ per hour sometimes
most of the gpu are around 0,200$
viral mason
YOOOOOOOOOOOOOOOOOOOOOO
all u need is two folders in your google drive for that, one called vocales and another one that can be named anything
knotty moth
ok moral of the story make it 1 hour
even 30 mins is also not bad (perhaps like 95% to 97% or something)
analog obsidian
#1419105823451517028 message i did this with 20 mins
fossil sage
even 30 mins is also not bad (perhaps like 95% to 97% or something)
bet
analog obsidian
so really what matters is how diverse and how good the dataset sounds
it has to cover multiple pitches
fossil sage
https://discord.com/channels/1159260121998827560/1419105823451517028/14191058234...
sounds pretty good
viral mason
then do this ⬇️
/content/drive/MyDrive/inputnoui (replace "inputnoui" with whatever the folder is u put the data into to clean
fossil sage
so really what matters is how diverse and how good the dataset sounds
is it voice to voice or tts
viral mason
@fossil sage
analog obsidian
is it voice to voice or tts
rvc model
fossil sage
then do this ⬇️
/content/drive/MyDrive/inputnoui (replace "inputnoui" with what...
alright ill check it out
viral mason
is it voice to voice or tts
there's sovit which is tts and rvc which is speech to speech
fossil sage
what you think of this audio
tawny radish
@fossil sage u can also dm me for any help with making a model
viral mason
same with me ^
tawny radish
what you think of this audio
its good
tawny radish
tbh
fossil sage
👍
tawny radish
just has his voice
viral mason
what you think of this audio
this doesn't sound bad at all
fossil sage
analog obsidian
for some datasets that are too clean i inject some white noise to help the model generalize
viral mason
uhh
fossil sage
get like 30mins
hallow thistle
Time matters, I sometimes have to open a UVR5 Colab notebook standby; when I'm ready I click "connect".
deft condor
what voice changer are you using? i used okada but it doesnt appear the mic in my discord lowk
hallow thistle
what voice changer are you using? i used okada but it doesnt appear the mic in m...
W-Okada.
!howtoask
analog obsidian
rvc models overtrain super fast, the official rvc docs recommend 20-40e, i personally prefer 40e-60e
viral mason
rvc models overtrain super fast, the official rvc docs recommend 20-40e, i perso...
that depends on dataset length tho or nah
analog obsidian
that depends on dataset length tho or nah
as long it's 10 mins or more, yea 40e is enough
viral mason
I just go based on the tensorboard and how good it sounds
analog obsidian
below than that no idea
viral mason
40 is usuallly never enough for any of my models
analog obsidian
I just go based on the tensorboard and how good it sounds
losses aren't accurate for voice models as the trend doesn't match with what are you hearing
viral mason
interesting
analog obsidian
so far they're useful when training pretrains, coz you have to monitor they do not diverge
which never happens with finetuning
tawny radish
i wish voicemodel commisions were still a thing
im too lazy to make my own voicemodels
viral mason
i wish voicemodel commisions were still a thing
me when u can just make them for free
tawny radish
me when u can just make them for free
im too lazy
analog obsidian
40 is usuallly never enough for any of my models
works for me in mainline, i remember in applio also 40e used to work just fine for me, even when using their messed up discriminator lol
viral mason
fair
tawny radish
when commisions were a thing i paid trhis one guy and he made some of my best voicemodels i still use today
i swear some commisioners in there were genuinley insanity
viral mason
when commisions were a thing i paid trhis one guy and he made some of my best vo...
what was their name?
viral mason
I don't recognize the name
knotty moth
im too lazy to make my own voicemodels
many model makers are using cloud/rental solution instead of having own capable gpu
tawny radish
many model makers are using cloud/rental solution instead of having own capable ...
ye ik he is
but
he makes really good voicemodels nonetheless
i used to reccomend him to everyone
hallow thistle
imcertibtw
I'm kind of familar with this name, I just don't remember where I saw this one.
deft condor
i use rtx 2060 whats best for me guys?
tawny radish
I'm kind of familar with this name, I just don't remember where I saw this one.
yeah im pretty sure a mod knew him
alot of mods
knotty moth
ye ik he is
I have made several models using kaggle for private use, and dont think im open to any requests
tawny radish
I have made several models using kaggle for private use, and dont think im open ...
is it really hard to make good voicemodels theese days?
i havent made one in a month or so
analog obsidian
you just need a good dataset and dont overtrain the model
rvc is still 2023 tech tho
tawny radish
is there ever new tech
knotty moth
is it really hard to make good voicemodels theese days?
hard to tell, it may sound okay for common ppl to hear, until there may possibly be minor artifacts and some static noise in the spectrogram
tawny radish
hard to tell, it may sound okay for common ppl to hear, until there may possibly...
see thats what i dont like
overtime people belive it less and less
that its real
native compass
why does the voice changer work when i put it on my cpu but when i put it on my gpu it dont work, does that mean my gpu is to old or some?
knotty moth
see thats what i dont like
I could use analogy with the modern game graphics: the native raster & raytraced rendering, and with DLSS/FSR. most ppl may find it okay while comparing with the hardware spec demand, and until spotting minor flaws like ghosting, shimmering, inaccurate shadows & lighting, etc.
tawny radish
I could use analogy with the modern game graphics: the native raster & raytraced...
hypathetically if RVC V3 was done being developed, would those issues even go away
or do they stay forever untill we get better ai technology in this type of field
fallow yacht
кто нибудь хелпануть может?
viral mason
this is an english only server Garrett Garrison
-# please get the reference
native compass
why does the voice changer work when i put it on my cpu but when i put it on my ...
no one?
viral mason
no help for yuo mr egril troller
native compass
no help for yuo mr egril troller
i quit on the egirl thing i swear 😭
this is for a normal voice changer
native compass
bru
native compass
why does the voice changer work when i put it on my cpu but when i put it on my ...
please someone help me 😭
tawny radish
i quit on the egirl thing i swear 😭
Ill help you
tawny radish
why does the voice changer work when i put it on my cpu but when i put it on my ...
What gpu do u have?
native compass
What gpu do u have?
nvidia quadro T1000
i think cuz it’s old
i’m on laptop not pc btw
native compass
😂
tawny radish
Dont even try
Wanna know something tho
I use w-okada for egirl trolling too 😭😭😭😭
Been doing it for years
viral mason
I'm so dissapointed in you
native compass
Dont even try
i figured 🤣
native compass
I use w-okada for egirl trolling too 😭😭😭😭
😂
viral mason
you're supposed to use it to play as characters not that yucky shit! 😭
tawny radish
you're supposed to use it to play as characters not that yucky shit! 😭
LMFAOOOOO
I dont do it as a kink thats wierd
Im gonna start making content with it
😭
Id blow up lwk
viral mason
what ever happened to using it to be Darth Vader not E Goth Latina Darth woman Vader
tawny radish
what ever happened to using it to be Darth Vader not E Goth Latina Darth woman V...
E girl latina
Old times dude..
I used to use that voicemodel too
😭
tawny radish
It was literally called egirl latina
I got funny reactions from people
Thats why I do it 😭
tawny radish
I don't trust like 99% of people who do it
Bro trust me ive been doing it since like rvc was even a thing
I only use it for the funny reactions
😭
carmine siren
viral mason
I only use it for the funny reactions
dawn pulsar
I use w-okada for egirl trolling too 😭😭😭😭
The original wokada?
viral mason
full imp
is weights broken? It won't let me upload the mp3
viral mason
no it just sucks really bad
serene pollen
there is any update on the OKADA MMVCServerSIO2 ?
it simply is not working anymore for some reason
viral mason
there is any update on the OKADA MMVCServerSIO2 ?
it simply is not working anymo...
what dose it look like, u might be using something outdated
viral mason
what even is that
teal ferry
<@498248765698867201>
They don't know what they're doing. You can train a model on a minute of audio if you want.
The foundation of my argument revolves around covering the entire phonemic spectrum. You may get something that sounds reasonable with less audio to some degree. But the model will struggle with under-represented phonemes in your dataset. So it will end up without nuanced emotional speech or sounds that sit on the extreme ends of a voice. The little tiny details that make a voice sound human will be absent.
fossil sage
what even is that
when using voice to voice when trying to clone voices it doesn't mimic the accent or the emotion to further explain think about it like this if you were to have the best model ever of a girl screaming and has a indian accent and if you to use rvc to try to mimic that it woudn't work due to does 2 factors
analog obsidian
rvc outputs are rather flat because the model is learning the spectograms
not the emotions
it takes a mel spectogram, trains it, then outputs a .wav
fossil sage
They don't know what they're doing. You can train a model on a minute of audio i...
oh ok makes sense just to clarify doh im not trying to make the best model ever or the best ai voice clone audio ever i am just trying to make it good enough and yes i know i sent this video multiple times but as you can see his audio is not the best either but its still good enough for viewers to enjoy the video and the audio
fossil sage
not the emotions
YES thats my point the reason why i cant do voice to voice is because number one im not good at reading out loud i mess up at bit and i mumble also im not good with accents and mimicing how some talks therefore tts is best
viral mason
It's basically a filter of a voice
In realtime you just gotta mimic how the character speaks
I've learned how to do that for the most part which makes models sound better in okada
analog obsidian
YES thats my point the reason why i cant do voice to voice is because number one...
some tts can learn expressions and emotions yeah, rvc cannot, it's just learning mel
novel pilot
I'm looking for an AI video to video tool where I can type in a prompt say "Ice Age" and then it converts the video with the characters and background into an ice age type enviorment and maybe the characters look like cavemen while keeping the same movements and audio as the original clip. Does anyone know one?
fossil sage
In realtime you just gotta mimic how the character speaks
yes but i am not using realtime
viral mason
yes but i am not using realtime
Just giving an example y'know
teal ferry
oh ok makes sense just to clarify doh im not trying to make the best model ever ...
That sounds horrible
fossil sage
That sounds horrible
yes it does sound horrible however it good enough to achieve 100k views or more on tiktok
fossil sage
mine sounds even worse
fossil sage
Is that a lot?
yes
teal ferry
The real question is will it get people to look for more of your content
It's not. I have YouTube videos with like five times that and they've earned nothing
viral mason
There's plenty of that stuff on YouTube and probably other video sites
fossil sage
It's not. I have YouTube videos with like five times that and they've earned not...
anyways this doesn't change the fact that ai voice i cloned was bad
fossil sage
mine sounds even worse
teal ferry
Imyoure trying to make useless content that will go into the void and stay there for eternity
It's a waste of time. Make something people want to consume and you will be successful
Make something that tries to find a loophole and you'll most likely fail. Unless you get really really lucky
fossil sage
Imyoure trying to make useless content that will go into the void and stay there...
??????
I'm confused
But anyways forget this
It don't matter
teal ferry
Also targeting tiktok as a platform is not a good idea. They pay like sht
fossil sage
OK i get what your trying to say
fossil sage
but as you see here it doesn't sound that good
would it be a good idea to run it through an model
rv
c
viral mason
What did u use to clean the audio? Is it that Collab I sent from before
teal ferry
Fail a ten thousand times
teal ferry
The ten thousand and first time you'll like what you hear
viral mason
teal ferry
Mastery is repetition. Stop looking for some magic recipe
There isn't one.
Just practice. Idiot
viral mason
That sounds good, how long is the dataset you made
teal ferry
I don't care
fossil sage
I don't care
viral mason
So I should baby him?
That's what you said not me
fossil sage
ist crazy how some people so easily are able to just use eleven labs and get high results
unfortunately its paid
viral mason
its not a model its inference
Wdym, I'm kinda slow lol
teal ferry
He can handle it. I yell at him all the time and he keeps coming back
.I respect that actually
viral mason
Alr u do u man
fossil sage
@viral mason
viral mason
ist crazy how some people so easily are able to just use eleven labs and get hig...
Tbh I don't like eleven labs due to it not being able to copy non human characters like Venom or General Grievous
At least I don't think it can
fossil sage
He can handle it. I yell at him all the time and he keeps coming back
.I respect...
your not yelling at me irl
teal ferry
True
But some people can't take direct criticism. You can
That's good. You will go far with rhat
viral mason
fossil sage
But some people can't take direct criticism. You can
well i guess it was constructive critism unlike when some people would just say quit you suck at this why are you even trying
teal ferry
Haha well there's a line between helpful and just being a jerk
Skin color is irrelevant also
Once you can make a bad model. Meticulously craft your dataset. And I mean painfully craft it its going to suck. But the result will be exceptional.
cursive thicket
i need help
fossil sage
Once you can make a bad model. Meticulously craft your dataset. And I mean painf...
ok but yk in the yt video i showed you he used vibe voice for tts then used rvc to make it sound better should i do that
teal ferry
ok but yk in the yt video i showed you he used vibe voice for tts then used rvc ...
You can sure. It all comes down to how well you train that RVC model.
But using RVC as a filter or enhancer in a sense is a good idea. The foundation of the sound though will have to come from the first model
fossil sage
You can sure. It all comes down to how well you train that RVC model.
what is a good ecpochs rannge for models
viral mason
i need help
With wha?
teal ferry
The amount of epochs depends entirely upon the size of the dataset and the model you're using
I cannot give you a number
Look at your training data output. Use something like wandb or tensorboard
This data is not for the faint of heart and will not really make any sense. There is also little content to learn about it on things like YouTube. Just stick with it though
viral mason
Tensorboard is the most common tool used to see training progress
teal ferry
Once you use wandb tensorboard looks like it's for kids
viral mason
Well what does it even look like
viral mason
What's that
cursive thicket
@viral mason
cursive thicket
but i need help
cursive thicket
already did all that but its js not going through mic and delyaed
fossil sage
The amount of epochs depends entirely upon the size of the dataset and the model...
whats a dataset
teal ferry
Well what does it even look like
theres way more but thats kind of the basic part of it
teal ferry
whats a dataset
a set of data
fossil sage
how long the audio is ?
teal ferry
no all of the audio
teal ferry
i havent used the voice changer. but i do know its extremely buggy
push buttons until it works
cursive thicket
Okay I will wait for @viral mason to help me
teal ferry
youll have to elaborate on your problem
all you said was you need help
its vague
what exactly is the problem youre seeing
royal kettle
what is a good ecpochs rannge for models
model is usually done around 40-100 epochs
teal ferry
vb audio is more confusing then it looks. i would suggest watching a video on hooking that up
cursive thicket
It’s delayed and isn’t working
teal ferry
yeah find a youtube video that specifically uses it with discord
i cant tell you off the top of my head what is wrong. but id assume it has to do with the virtual audio cable and pass through with discord
cursive thicket
Im see if @viral mason if not idk what Im gonna do
fossil sage
model is usually done around 40-100 epochs
lets hear yo modesl
royal kettle
lets hear yo modesl
?
royal kettle
voice to voice
royal kettle
Not really
fossil sage
man its even better then elevens models
fossil sage
Not really
how long was the wav file when u trained it
royal kettle
how long was the wav file when u trained it
1 hour 47 minutes
royal kettle
why does it sound like this btw
Because the model was trained with the SPIN embedder and when weights does its inference it doesnt use spin but uses cvec
fossil sage
Because the model was trained with the SPIN embedder and when weights does its i...
well idk what this but imma just assume thats it going to sound ok when i put it through rvc
royal kettle
as long as you select the spin embedder you will be fine
fossil sage
do you use appilo
royal kettle
Yea
cursive thicket
So can one of y’all help me
serene pollen
what dose it look like, u might be using something outdated
i can select any input/output but it doesnt "comes out"
nice, just cuz i came here and asked for help, now it is working
sorry for bothering
serene pollen
why is ur imput vb audiocable?
ignore that, i know isnt the best for it but tahts what i use
tawny radish
ur imput is supposed to be ur main microphone
serene pollen
and ur output is ur speakers?
my output is my speakers cuz i was testing
serene pollen
also, it stopped again 🤡
tawny radish
my output is my speakers cuz i was testing
monitor does that
u can listen to urself with monitor
tawny radish
also, it stopped again 🤡
try this
switch to server instead of client
use WASAPI, youll know what im talking about once u click ur devices
for the number thing on server, put 4800
i think
or
48000
yes
btw
ur extra is broken
cursive thicket
i know ur intentions buddy
So can you help me
cursive thicket
@tawny radish
viral mason
I got 6 pings what'd I miss, I was showering
viral mason
Maybe they should shower too
tawny radish
I got 6 pings what'd I miss, I was showering
bro ive been in this server since febuary i never seen u speak here til now wth
like
last year febuary
@fading lodge is my old account
viral mason
I've been more active lately other than just posting models
Actually interacting with ppl because why not
viral mason
<@1151595680675139686> is my old account
Did u have me added on that acc btw?
viral mason
Ah
I'm gonna scroll up to see what silly people needed my help
And then not help them
fossil sage
And then not help them
knotty moth
it doesn't even look like rvc models, you shouldn't recommend such strange things for rvc model makers & users, even from sketchy youtube videos
viral mason
Looks like Tensorboard if they locked in
viral mason
<:pepe_stare:1159360351872221274>
Well I'll help anyone I was helping before lol
Just joking around
steady inlet
how good is vonovox ? i have a rtx 3060 with an amd ryzen 5 5600g but with w okada tg i still crackle
can it be because of the model ?
tawny radish
how good is vonovox ? i have a rtx 3060 with an amd ryzen 5 5600g but with w oka...
i could help you fix the crackling if u want
vonovox just gets updates alot id use it if i were you
it just has a BIT more delay
steady inlet
it's so annoying!! i thought my computer was good enough :./
tawny radish
which version do you use?
steady inlet
just want non-crackling voice that sounds robotic at the end of my sentence
I'm downloading vonovox now, I was using the latest TG fork of deiteris
tawny radish
I'm downloading vonovox now, I was using the latest TG fork of deiteris
do u want me to help you set it up?
i can give you proper settings
im actually new to using vonovox and its easy
steady inlet
yeah would be awesome ngl
steady inlet
because
tawny radish
just dm me
steady inlet
ok
steady inlet
just dm me
did
teal ferry
it doesn't even look like rvc models, you shouldn't recommend such strange thing...
thats not rvc correct.
its a way to read data.
you can use whatever you want i dont give a fuck
one tool will give you more control over your data and the other one wont. what a beginner does or doesnt use is irrelevant because they will have no idea what theyre looking at anyways. The entire point was to just try.
viral mason
i had minimal to no delay
what's the best settings? I plan on switching over once it's updated to finally have more than 8 slots
tawny radish
what's the best settings? I plan on switching over once it's updated to finally ...
vonovox?
or
w-okada forked
viral mason
vonovox
tawny radish
i can send u mine
depends on ur gpu tho
mine dosent have TOO much delay
it has no crackling
viral mason
didn't I give those settings to u or someon?
tawny radish
hmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmm
viral mason
my gpu is this btw
tawny radish
i frogot
viral mason
really?
viral mason
I always thought it was kinda mid since the numder was lower than 3000 smth
viral mason
is that one better than mine
tawny radish
yes
viral mason
I might upgrade eventually
knotty moth
prob better then a 2070
gaming/raw performance wise, probably yes but the RTX one would outperform it in AI tasks due to usage of tensor cores
balmy junco
Anyone know of ways to match accent ? I tried index file and tuning the index meter, not much effect.
teal ferry
<@498248765698867201>
couple things. youre not in the virtual environment and you need to be in the same folder as the whl
so first activate the venv. then move the .whl into the folder the terminal is in or cd the terminal into the folder the .whl is in
or provide pip with the absolute path
oh you see how you installed it it says installed but then alltalk cant see it
its because you installed outside the venv. so you need to
fossil sage
like this
C:\Users\yonshuk>dir
Volume in drive C is FSOS
Volume Serial Number is A42B-E89B
Directory of C:\Users\yonshuk
09/21/2025 08:02 PM <DIR> .
05/16/2025 07:11 PM <DIR> ..
09/21/2025 07:18 PM 417 .bash_history
09/18/2025 06:52 PM <DIR> .cache
09/21/2025 03:06 PM <DIR> .conda
09/20/2025 01:29 PM <DIR> .dotnet
09/17/2025 06:41 PM 126 .gitconfig
09/17/2025 06:44 PM <DIR> .local
09/18/2025 07:02 PM <DIR> .matplotlib
09/21/2025 07:51 PM 227 .python_history
09/06/2025 12:49 PM <DIR> .stacher
09/18/2025 05:01 PM <DIR> .venv
09/18/2025 06:07 PM 0 1.0.4
09/17/2025 09:25 PM 0 App
09/17/2025 09:36 PM <DIR> chatterbox
09/17/2025 09:33 PM <DIR> chatterbox_old
05/16/2025 07:11 PM <DIR> Contacts
09/21/2025 05:34 PM <DIR> Desktop
09/21/2025 08:02 PM 0 dir
09/21/2025 03:04 PM <DIR> Documents
09/21/2025 08:16 PM <DIR> Downloads
05/16/2025 07:11 PM <DIR> Favorites
09/20/2025 11:13 AM <DIR> ffmpeg
09/02/2025 05:28 PM 0 ffmpeg2pass-0.log
09/18/2025 05:38 PM <DIR> index-tts
05/16/2025 07:11 PM <DIR> Links
09/21/2025 03:07 PM <DIR> miniconda3
09/10/2025 08:35 PM <DIR> Music
07/13/2025 08:19 PM <DIR> Pictures
05/24/2025 01:22 PM <DIR> Saved Games
05/18/2025 08:22 PM <DIR> Searches
05/22/2025 04:21 PM <DIR> Superposition
09/21/2025 07:52 PM <DIR> venvs
09/21/2025 08:10 PM <DIR> Videos
7 File(s) 770 bytes
27 Dir(s) 197,643,468,800 bytes free
C:\Users\yonshuk>
teal ferry
no the project root
so alltalk is the project
so the projects main folder is another way to say it
most likely its called alltalk_tts
here
cd e: OR e: (depends if youre in cmd or wt)
cd xtts\alltalk_tts
dir
fossil sage
here
Directory of E:\xtts\alltalk_tts
09/21/2025 07:40 PM <DIR> .
09/20/2025 12:03 PM <DIR> ..
09/20/2025 12:03 PM <DIR> .github
09/20/2025 12:03 PM 110 .gitignore
09/20/2025 12:03 PM <DIR> .vscode
09/21/2025 07:40 PM 5,793 2.2.2+cu121
09/21/2025 07:20 PM <DIR> alltalk_environment
09/20/2025 12:03 PM 22,932 atsetup.bat
09/20/2025 12:03 PM 18,415 atsetup.sh
09/20/2025 12:11 PM 0 cmd_windows.bat
09/21/2025 08:06 PM 700 confignew.json
09/20/2025 12:03 PM 15,338 diagnostics.py
09/20/2025 12:03 PM 553 docker-compose-cuda.yml
09/20/2025 12:03 PM 749 docker-compose.yml
09/20/2025 12:03 PM 818 dockerconfig.json
09/20/2025 12:03 PM 818 Dockerfile
09/20/2025 12:03 PM <DIR> finetune
09/20/2025 12:03 PM 104,941 finetune.py
09/20/2025 12:03 PM 142 launch.sh
09/20/2025 12:03 PM 35,184 LICENSE
09/20/2025 12:03 PM 941 modeldownload.json
09/20/2025 12:03 PM 9,258 modeldownload.py
09/20/2025 12:44 PM <DIR> models
09/20/2025 12:03 PM 828 nvidia.Dockerfile
09/20/2025 12:46 PM <DIR> outputs
09/20/2025 12:03 PM 111,877 README.md
09/20/2025 12:03 PM 47,659 script.py
09/20/2025 12:25 PM 332 start_alltalk.bat
09/20/2025 12:25 PM 308 start_environment.bat
09/20/2025 12:25 PM 334 start_finetune.bat
09/20/2025 12:03 PM <DIR> system
09/20/2025 12:03 PM 57,861 tts_server.py
09/20/2025 12:03 PM <DIR> voices
09/20/2025 01:03 PM <DIR> pycache
23 File(s) 435,891 bytes
11 Dir(s) 205,143,212,032 bytes free
E:\xtts\alltalk_tts>
teal ferry
double click start_environment.bat
fossil sage
just did
teal ferry
that should open the terminal with the env activated then pip install the whl
fossil sage
teal ferry
right click in that folder and open in terminal
or in the address bar of explorer type cmd hit enter
tell me which one you did so i know if youre in cmd or windows terminal
if you right clicked and went to open in terminal then type
./start_environment.bat
if you typed cmd in the address bar do
start_environment.bat
fossil sage
im using cmd
teal ferry
if it closes again on you then right click the bat file open it with notepad and at the very very bottom i want you to put this word
pause
file then save and then try to run it from the terminal again
it should stay open now and then you can tell me the traceback
fossil sage
i opened it and its blank
fossil sage
yes nothing in it
which was why it wasn't working
looks like i should reinstall it
teal ferry
ok scroll up and you see a folder alltalk_environment go in there
you can reinstall it but i mean we dont need it if we know what were doing
in alltalk_environment folder you should see like an env folder. open that, then scroll down you should see a python.exe right?
sorry python.exe should be in env/bin/python.exe
fossil sage
sorry python.exe should be in env/bin/python.exe
theres no bin folder
teal ferry
im working off memory here.
bin could be linux only whats in the env folder
were looking for python.exe
search the env folder for it if you have to
fossil sage
were looking for python.exe
i found it
E:\xtts\alltalk_tts\alltalk_environment\env
python.exe
teal ferry
ok go back to the terminal and type
E:\xtts\alltalk_tts\alltalk_environment\env\python.exe -m pip install put\the\path\to\deepspeed\whl\here
just copy and paste all of that delete the last part after "install" and put the whl there
fossil sage
ok go back to the terminal and type
```
E:\xtts\alltalk_tts\alltalk_environment\...
C:\Users\yonshuk>E:\xtts\alltalk_tts\alltalk_environment\env\python.exe -m pip install C:\Users\yonshuk\Downloads/deepspeed-0.14.0+ce78a63-cp311-cp311-win_amd64.whl
Processing c:\users\yonshuk\downloads\deepspeed-0.14.0+ce78a63-cp311-cp311-win_amd64.whl
Requirement already satisfied: hjson in e:\xtts\alltalk_tts\alltalk_environment\env\lib\site-packages (from deepspeed==0.14.0+ce78a63) (3.1.0)
Requirement already satisfied: ninja in e:\xtts\alltalk_tts\alltalk_environment\env\lib\site-packages (from deepspeed==0.14.0+ce78a63) (1.13.0)
deepspeed is already installed with the same version as the provided wheel. Use --force-reinstall to force an installation of the wheel.
this already happened to me twice
full imp
no it just sucks really bad
so there's no way to use it?
teal ferry
E:\xtts\alltalk_tts\alltalk_environment\env\python.exe -m pip install deepspeed-0.14.0+ce78a63-cp311-cp311-win_amd64.whl --force-reinstall
also though why are you installing this
you do not need it for the repo/project to function correctly
given you degree of experience and the fact i have to debug through you. Its a waste of time. While we could get it working, you should ask yourself first if you need slightly faster inference
actually i can answer for you. You dont need this right now. But, if youre dumb and want to continue fixxing it then i need to know the traceback or error that youre getting that caused you to attempt to install this
fossil sage
given you degree of experience and the fact i have to debug through you. Its a w...
i've been using chatgpt for help and it put me through a infinite rabbit whole
teal ferry
well you have it installed
so im not sure what caused you to continue to try and install it
fossil sage
so im not sure what caused you to continue to try and install it
its because alltalk said it wasn't installed
teal ferry
oh, ok so we need to debug alltalk source code
you can pay me to do that. its $210 an hour
or
ignore it
fossil sage
you can pay me to do that. its $210 an hour
💀
teal ferry
you dont need it
theres a small chance he exits the loop if deepspeed isnt detected. but im going to guess he didnt code that
fossil sage
also does alltalk tts use inference and model at the same time
fossil sage
and second i said ignore it
ive tried generating it with the error
so
i did listen to your secon step
anyways
teal ferry
youre opening up settings and docs
there should be another url in the terminal
i dont remember which port he opens up but its not the one youre using now - 7851
try 7852
fossil sage
well i gotta go to bed soon but uh somehow i used chatterbox then i used appoilo to convert the audio and guess how it sound
teal ferry
fossil sage
teal ferry
just click that link
fossil sage
just click that link
ok
teal ferry
fossil sage
http://127.0.0.1:7852?__theme=dark
great its not working