#✨│ai-help

1 messages · Page 327 of 1

low shard
#

btw that's the Beta, it's not stable but it should work fine

#

the settings are explained in the docs

magic hare
#

gotcha, ty

magic hare
#

Idk if my settings are good, I'm still struggling with delay.

image.png image.png
karmic trellis
#

hello if the model i found and use did not have an index file included, can i use an index file from another voice i downloaded? can the two work well together?

low shard
magic hare
#

bro???

magic hare
proven hill
#

<@&1159293140440723499>

viscid moss
#

Weirdo 😭

viral mason
viral mason
low shard
magic hare
low shard
# magic hare yeah, sure

F0 method: fcpe
AP BWE 48K Upscaler: off
Block Size: 0.10
Close every other program in the background

but, you can't get "0 delay", that doesn't exist, 1 second was already good enough

supple trail
#

clean curve is more robotic sounding becausing when sighing its mostly just "white nosie" (air) on a graph this looks very fuzzy broken line with a bunch of random dots
screaming creates overtones and distortion. the pitch tracker sees a 5 or 6 different lines at once and doesnt know which one to follow
humming often have "vibrato" (shaking) or very low frequencies that drop off the bottom of the map

so basically the reason why clean curves make u sound like a robot

Standard pitch trackers (like the older PM or Harvest methods) are programmed to force your voice into a "clean curve" by deleting the "messy" parts.
It sees the air in your sigh as "trash" and deletes it
Result: Silence or a robotic "click."
It sees the raw energy of your scream as a "glitch" and tries to flatten it into a steady note
Result: The "Reading a Book" sound.

^^ RMVPE
FCPE doesnt try to force a clean curve it is context aware meaning it looks at the "messiness" around the sound and realizes its not a glitch its a human being breathing so therefore it maps the "messy" data more accurately so the AI knows how to produce a "messy" (human) output instead of a "clean" robotic one.
^^ FCPE

@viral mason (lots of thinking btw)

viral mason
supple trail
#

np.

#

im a previous user here before

viral mason
#

I will do some testing tho to see what I like

supple trail
#

@kindred surge , @deft pewter

supple trail
viral mason
#

what happened to those two accounts?

supple trail
#

discord disabled them

viral mason
supple trail
#

first account lost my 2fa since someone stole my phone, second account got banned aka suspended due to a server i haven't touched which i forgot i had a bot that i made which deletes alot of channels and recreates channels that have messages sent in etc, basically a server renovator which u can use to code a whole server with i just added hella shortcuts and improvements so i was able to optimize a server that is worth 11k lines up to being the same output just instead 100 lines

#

of code

#

got banned because of that stupid reason which had no other members other than me and the bot lol

#

you can friend me if you'd like idm and anyways i have to sleep, good night!

viral mason
supple trail
#

nice game favorite xd.

#

-# mines Death Note: Killer Within

#

anyways srs gtg to sleep.

viral mason
#

night night

viral mason
craggy brook
#

Is there anyone who can do this? By sharing it as a request.

cursive cypress
#

I have a question. I am trying to use TTS Voice Wizard and Okada together. I managed to get TTS Voice Wizard and Okada to work, but my last remaining problem is routing the output of Okada to Discord. The easiest solution is to use a separate virtual cable... but as far as I am aware you can only get 1 virtual cable through VB-Audio... unless you pay for more. But I've heard you can use VoiceMeeter Potato's Virtual Input/Output to route the audio from one application to another, but I have no clue how to do that. So, I am here. Wondering if anyone can help me with my current predicament.

Edit: Oh I almost forgot.
*AMD Radeon RX 5600M (surprisingly not struggling as much as I thought it would)
*Windows 10

finite wind
#

has anyone explained the difference between RMVPE and RMVPE+ in here?

#

discord search eefs up the + part so I can't search

#

aihub docs doesn't explain it in detail

viral mason
#

I don't know if that's a thing

viral mason
#

I tried voicemeeter + vac lite + vb cable

#

and also having fl studio and voicemod

#

nothin

cursive cypress
#

Maybe I will have to pay for a separate virtual cable after all... I mean, it should work, right?

viral mason
#

I wouldn't do that

cursive cypress
viral mason
#

try asking a helper maybe like the one online that I cannot pronounce their name

#

just ping the role they should be able to try

cursive cypress
#

I just wanted to use your GlaDOS voice model alongside TTS. Why does the world conspire against my simple wish cat_angry

viral mason
#

I just use it in realtime with a autotune effect from fl studio

cursive cypress
#

That's the easiest and most correct way of using it, yes. But I've gone through all this chaos to make the TTS sound like GLaDOS lmao. (Actually do you know any other TTS application that can work alongside the voice models?) If there isn't, that's fine. I guess its finally time I give up being a coward and actually use my voice cat_gato... which would probably be a good chance of pace

viral mason
#

sadly I don't know any software that can do tts and connect to wokada/vonovox

#

example of it in realtime

cursive cypress
#

You might just've convinced me to use VC normally. Thanks

viral mason
#

lol you're welcome

#

btw if u want the plugins I used the autotune is kinda pricey but worth it for glados

#

gform is free

cursive cypress
#

I will keep that in mind. This is my first time messing with this kind of stuff, if I end up enjoying it more than I expect, I might just buy the plugins.

viral mason
#

it's well worth it tbh, tho keep in mind the setup requires Fl studio, Voicemod, Voice meeter (I use banana) Vb cable, and Vac lite

#

all of that stuff is free tho

cursive cypress
#

Well, let me use the snipping tool real quick. I will be saving this for later. Thanks :3

viral mason
#

you're very welcome!

#

you can add me if you want and I will gladly help you ^^

#

btw what gpu do u have

#

for realtime AMD can use Wokada tg fork, and Nvidia can use Vonovox and tg fork

#

vonovox is what I used in that example

cursive cypress
#

Oh like I said it's a AMD Radeon RX 5600M. The thing is still working somehow. I should upgrade the setup pretty soon. I bought this ages ago when I was a little too short on cash

viral mason
#

whenever u get the money I suggest anything like a Nvidia 3060 or higher

cursive cypress
#

(If I knew all the tinkering this rusty thing needed from the start, I would never have bought it lmao!!!)

viral mason
#

lol

cursive cypress
#

but then again the last time i checked GPU prices was ages ago

craggy brook
dim cape
#

i occasionally come here to train some rvc ai voice model, can i know what's the newest collab to train models or where to? and if i have to use a different pretrain for spanish talking voice models?

#

im training with a dataset of 2 minutes on spanish

#

normal talking model no singing
oh yeah and id like to know what kaggle notebook you use for training

cursive cypress
#

well, managed to make TTS and okada work on discord. Sounds as awful as you'd imagine. I am really better off using my own voice from now own

low shard
low shard
low shard
cursive cypress
#

oh wait google exists for a reason

low shard
cursive cypress
low shard
#

you could check the guide to know more

cursive cypress
hallow thistle
cursive cypress
midnight fern
#

How do i make ai covers?

finite wind
#

can I use spinv2 for 48khz dataset? seems like there are no chat logs for this

brazen dove
#

whenever i speak sometimes its like shaky, like it just goes low kinda robotic-y, i tried changing settings but its still the same, could someone maybe help config?
RTX 4070
windows 64x
trying to use the voicechanger

simple ore
#

as far as I know

finite wind
#

but thanks a lot

#

I just gotta try both legacy 2.9 and og pretrain for 48k and see which is better

#

for realtime and TTS not singing

hardy yew
#

which is far from perfect

foggy ore
#

Getting claude to give higher detailed work and higher understanding abilities

finite wind
hardy yew
#

the art of training a pretrain from scratch as good as OG remains a mystery

finite wind
#

yeeeeep

hardy yew
#

some people like Lyery have done a lot of work in this direction, but eventually the OG just remains better for some reasons

finite wind
#

I just gotta see myself how awful of 48k og pretrain is as others say

#

so it's uh

foggy ore
#

Ignoring me😔

finite wind
#

og 48k pretrain, legacy 2.9 and maybe 1.5 but I kind of know 1.5 is solely consist of singing data not talking

hardy yew
finite wind
#

it's like I'm in the field of expertise where we lost our technologies in the dark ages

#

not in a literal sense but yk

hardy yew
finite wind
#

and even though 2.9 is having issues with cracking, I saw some people recommending it for realtime talking models and TTS

#

that's why I decided to try it out

hardy yew
finite wind
#

oh the og one uh

hardy yew
#

but personally I didn't experience the vibrato issue

finite wind
#

48k og somehow have a bad rep

hardy yew
#

and training on singing dataset doesn't mean the pretrain is bad for normal speech

#

it should do fine

hardy yew
finite wind
hardy yew
#

yeah, that's always a nice thing to do

#

actually, for various data, different pretrains might turn out better

finite wind
hardy yew
#

As i said, usually with training people go for either OG or 1.5. And sometimes one is superior, other times the other one

hardy yew
finite wind
finite wind
hardy yew
hardy yew
#

I mean only singing tests like that, but in there vibrato is usually already present in the input sample

#

I guess i should just record myself holding a long note

finite wind
#

oh just one more thing before I go, I personally experienced that 32k have clear audio quality decrease compared to 48k models when talking, this has been the same case for my friends to audit via discord claimed 48k sounds quote "almost like as if you changed your mic to a more high quality and expensive one"

finite wind
#

maybe we'll find something interesting

hardy yew
#

But not sure if that's what could instantly be perceived by people over discord as "higher quality mic"

finite wind
#

eh one of them have crazy equipments so I'd assume that could be the reason

hardy yew
#

I would expect it to be more of a "different sound characteristics" case than simply a wider spectrum

#

but that's just a guess

finite wind
#

I don't think 48k has a clear improvement over 32k in terms of audio quality THAT much

hardy yew
#

From what I heard, sybilants and breath noises are more easily going bad in 48k compared to 32k

#

I think that was one of the reasons 32k is recommended over 48k in most cases

#

But I can't speak from experience with 48k because there's almost none

finite wind
#

trying like ship, chip, keep, steep then say something without sybilants

#

huge artifact difference

#

could be the dataset problem but at least I do experience that with 48k model

hardy yew
#

I guess there is a chance that some dataset preprocessing would help

finite wind
#

but then I realize this whole RVC thing doesn't handle high frequency that well in the first place

#

we even numb some high frequency poking out in the dataset a lil too

hardy yew
finite wind
#

so I guess sibilants could be naturally hard or even impossible to make naturally in any model

hardy yew
#

yeah, i think the difficulties of 48k over 32k might be a result of literally that. Wider frequency spectrum, more room for high frequency details which can be difficult to replicate properly

#

(FTR this is just a speculation as well, someone with better experience here could be able to tell for sure)

finite wind
#

THEN something goes wrong then I legit don't know what to fix from the dataset at that point

hardy yew
#

and then splitting into short clips (3s i think?) is done as preprocessing in Applio

#

so that's two of the things you described done more or less automatically

#

(definitely not perfect, obviously)

finite wind
hardy yew
#

understandable

#

the smartcutter was kinda made specifically for this purpose

#

so ideally it should keep breaths and remove dirty silence

#

tested it a bit when it was in development and it looked nice

finite wind
#

I'll definitely try it out as well

#

compare it to my manual work and hey, if it's doing better job than me

#

fine by me

gleaming crescent
#

Is selling ai tool allowed?

finite wind
#

maybe I'll upload a British accent realtime model when I'm done with this too

#

could be a far stretch but oh welp

gleaming crescent
hardy yew
# finite wind I'll definitely try it out as well

there's also a debug flag in the script that inserts noise into where it detected silence. That way you can compare side-by-side the source sample to a processed one and see what was and what wasn't considered silence.
Gives a nice overview of how it performs with your data

hardy yew
finite wind
#

actually it could go up to the AIhub as well if it's really nice

hallow thistle
hallow thistle
foggy ore
hallow thistle
foggy ore
#

Yeah i mean right now im geting claude to generate its own plan from the trending datas and perfornalism of a high quality working studio team for claude code so it should have the context needed

foggy ore
green quail
#

what can i use to make good rvc v2 models for free
(i need them for Replay bc weights.gg shut down)

hallow thistle
supple trail
#

question is using spin v2 better than content?

#

im hearing that using spin v2 is better when combined with fcpe

simple ore
#

there's no good spin v2 pretrain, the one I trained from scrach using refinegan is unfortunately not good for realtime and stuff

#

it needs more data as per dr87

supple trail
#

i heard spin v2 is pitch invariant tho

simple ore
#

what

supple trail
#

like something its better if you dont sound the exact accent of what the person ur trying to replicate sounds

#

im just trying to say i did search it up and found some ppl who used it say that spin v2 is good for sounding exactly like the model voice?

#

both for model training and tg devs fork owkada

#

but im not 100% sure

#

so i want someones confirmation

hallow thistle
supple trail
#

fcpe for my experience delivers almost the same accuracy as rmvpe and js faster

#

also with more benefits?

simple ore
#

it may be better for realtime

#

where you sacrifice a bit of precision vs speed

supple trail
supple trail
#

but yeah

#

i'd like to be able to have my screaming, humming, laughing, etc. sound better

low shard
low shard
low shard
low shard
supple trail
#

yeah i know

hollow hatch
#

I’m looking for the best AI subscription to use daily for things like random questions, analyzing situations and decisions, & helping me grow a business. I’m also somewhat interested in AI video tools like Sora or Veo but thats not a priority. Any recommendations on the best subscription?

hallow thistle
supple trail
#

slight visual bug but eh

hallow thistle
marble condor
#

What is the best rvc voicechanger

hallow thistle
#

What is your PC GPU? And what will you use the voice changer for?

full sky
#

:V Does anyone know how to make a RVC Voice Model? Cause I want to make Ai Covers with them

hallow thistle
full sky
#

I'm uhh...on mobile lol.

#

But I do have a laptop

hallow thistle
full sky
marble condor
#

Windows

hallow thistle
full sky
hallow thistle
marble condor
#

Of the voice changer

#

I just know about the w-okoda one

hallow thistle
marble condor
full sky
marble condor
#

Like what is the delay of it

marble condor
#

Like how much

full sky
hallow thistle
marble condor
marble condor
hallow thistle
#

My 2012 laptop doesn't have any GPU appear in Task Manager, so not ideal to run any voice changer. nso_ame_shrug

marble condor
#

Imma first try w okoda out and then vonovox

#

Thx

hallow thistle
#

But even if my laptop had a dedicated GPU, it probably might be one of those NVIDIA GeForce GT 600 series, which is also old and won't gonna work with any AI program.

hallow thistle
marble condor
#

Have a great day

marble condor
#

What happend

#

I joined like 3 days ago

hallow thistle
marble condor
#

K ty

lime acorn
#

where i can download RVC?

hallow thistle
# lime acorn where i can download RVC?

Don't try to ask something simple. What is your PC GPU? Realtime or non-realtime? And what will you use the program for? This is how I ask people here every time.

supple trail
#

im assuming their probably here for ai rvc real time voice changer.

#

considering they joined the server last month + the fact their playing a roleplaying server.

hallow thistle
#

Well, that sounds a bit awkward. MikaCakeRoll

supple trail
#

why awkward?

lime acorn
#

But I formatted the PC and I can't get the program

hallow thistle
lime acorn
#

🤨

hallow thistle
#

RVC (retrieval-based voice conversion) doesn't always mean realtime voice changer, at least be specific.

daring bough
#

Does sm1 have a Vox Hazbin Hotel english RVC model here

#

Bc all the old ones got taken down 😅

hallow thistle
daring bough
hallow thistle
supple trail
#

-rvc

low shard
low shard
supple trail
#

yh i know

#

but i searched on her profile

#

shes playing a gta in a roleplay server

#

also she included "used to test voices" shows she prob means about real time voice changer.

# 
|=1. i suggest you to use cloud based servers, your gpu isn't gonna handle the voice changer really well when running heavy games so i suggest you use cloud based servers instead.
|
|-2. if your new to using voice changers i suggest you use "Tg Develop's W Okada Fork"
|-|
|-|-2.1. if you want to train a model instead + with a voice changer embedded in it use "Applio"
|
|-3. if its laggy close any other tab and leave only kaggle and the main tab, everything else just close it.
|-|
|-|-3.1. if it sounds choppy go to task manager and find the tab that holds the main tab and set its priority to high ( don't ask why it just works im too lazy to explain )
#

@lime acorn

low shard
supple trail
#

i mean hey

#

if i made a mistake ill still have you guys to correct me

#

xd

supple trail
low shard
supple trail
#

fire

supple trail
# supple trail -rvc

by the way i assume you know alot about the bot cmds right? can you possibly tell me about them all

low shard
patent trellisBOT
# low shard -prefix-commands
🗣️ All Prefixes Commands

Commands that give you useful information that work with the prefixes (- or !)

-prefix-commands

Shows this list

-rvc

Shows RVC Documentations

-audio

Guides for Audio Cleaning and making Datasets

-colab | -googlecolab

A list of Useful Google Colabs Notebooks (Cloud)

-kaggle

A list of Kaggle Notebooks (Cloud)

-huggingface | -huggingfacespaces | -hf | -spaces

A list of HuggingFace Spaces (Cloud)

-lightning | -lightningai

A list of Lightning AI Notebooks (Cloud)

-realtime | -wokada | -rt

A list of Programs for using RVC Models in Realtime for Calls/Games (mostly Wokada & Wokada Deiteris fork) with Guides

-uvr | -ultimatevocalremover

Shows useful links about UVR, a program for vocal and instrumental separation

-howtoask

Shows how to ask properly for help

-howtosearch | -searchmodels | models

Shows how to search RVC AI Voice Models

-tracker

Shows a very old Google Spreadsheet with old RVC Models, not much suggested though

-gui | -rvcgui | -rvceasygui

Shows others that RVC Easy GUI is Outdated

-svc | -so-vits-svc

Explains others that So-VITS-SVC is very outdated

-javavsjavascript | -java | -javascript | -js

A text explaining the difference between Java & JavaScript (for fun)

supple trail
#

tysm

#

ill be sure to use it when ppl need help

low shard
#

oh yeah you were an old member

lime acorn
lime acorn
torn orbit
#

-realtime

patent trellisBOT
# torn orbit -realtime
🔊 Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

low shard
low shard
torn orbit
lime acorn
low shard
low shard
lime acorn
low shard
patent trellisBOT
# low shard -realtime
🔊 Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

supple trail
final canopy
#

hey guys, i just joined this server. think its a good server to learn stuffs about AI being an absolute rookie?

brisk wharf
#

hey so im tryna setup okada on my gfs pc, and i adjusted the extra and chunk akkordingly to what it needs while gaming in order to still work, but for some reason after the pc has been on for a little the voices will somehow always sound like a microwave no matter what we do with the pitch, the chunk is in a very wide margin of range so it doenst stutter, the voice itself doesnt stutter but will never sound like the model at all in game, even when i test the exact same model on my own pc, where it DOES sound good, she has a 3060

supple trail
#

tg fork
or det fork
or just owkada

#

wokada*

brisk wharf
#

tg-develop btw

supple trail
#

does it sound

#

like choppy overtime after using it?

#

are you running it cloud or locally.

brisk wharf
#

tried both client and server doenst rly make a difference, changing it to regular rmvpe instead of the rmvpe(onnx) it is almost completely solved

#

the voice would run smoothly but sound completely weird

supple trail
#

try out fcpe onnx

brisk wharf
#

like out of the pitch range

supple trail
#

are u using it cloud or

#

local

#

RN

brisk wharf
#

client rn

#

not server

supple trail
#

alright go to task manager

brisk wharf
#

mhm

supple trail
#

you should find something starting with an "M"

#

mmvc server or smt

brisk wharf
#

ye

supple trail
#

switch its priority to high

brisk wharf
#

oke

#

did that

supple trail
#

allows the pc to focus more resources on the voice changer.

#

alright try using it now

#

if you still notice choppiness

#

wait there is 2 differences

#

start delay vs choppy

brisk wharf
#

ye no it wasnt chopping/cutting out, it was just purely sounding off like a model would when u have the pitch on like -30

supple trail
#

which one? (start delay ignores like the first few milliseconds or seconds of u talking then start changing the later recorded voice)

supple trail
#

screenshot the model settings?

brisk wharf
#

so for context, she wanted to use a roxanne from fnaf model, she is a female, so 0 pitch would work fine on that, like on my pc it worked for her voice, we move over to her pc, and the same 0 pitch sounds completely off, and by off i mean if u were to imagine going on my pc and cranging the models pitch down to -40 far outside the models range, where itll sound like a dying microwave, even tho the pitch and format and index were the exact same as on my setup, on her pc

#

everything at 0

#

her chunk size is 480 since it wont work on lower delay while playing a game

#

and 2.7 extra

winged adder
#

How can i use whisper remotely but using the Microphone on the client ?

low shard
winged adder
twilit dagger
#

Hi Everyone, I have a question. I am looking for an AI bot to switch to. Currently we use ChatGPT 5.4 at our company. We have spent 3 years training in in our thought process and writing. We want to be able to give employees access without rewriting or changing its memory without approval.

So we are looking for a chatbot that can do 3 things.

  1. Move the memory we have developed in ChatGPT over to the new model.
  2. Allow employees to have permission to ask questions, but anytime memory is tryed to change, it will not change unless it is aproved by the managing account.
  3. We want multiple people to be able to use the memory and chat at the same time.
  4. We also need it to be able to scale, as we grow fast.

Any advice?

viral mason
glass smelt
#

how do i set up the cloud server, or is there a way to make it so local isnt extremely choppy and laggy and delayed

low shard
viral mason
#

<@&1159293140440723499>

dire portal
#

guys what are the best voices???

midnight fern
#

How do i make something sound more like the ai model than rely on the audio?

#

Bc its like 70% accurate

dense blaze
#

Can somebody please help me set up

winged adder
#

Can someone tell me if the following config Is accurate ?

This is a sophisticated "Remote GPU Pipeline" architecture. You are essentially offloading the heavy lifting of AI transcription from a mobile device to a high-performance home workstation. By using Buzz (a GUI for OpenAI's Whisper) and Parsec for low-latency streaming, you’ve created a real-time transcription bridge that bypasses the hardware limitations of a smartphone.

  • Parsec Requirements: The Home PC (Host) must have the Parsec Virtual USB/Mic Driver installed. This allows the PC to recognize the incoming audio stream from Phone 1 as a local microphone input.

Here is how it compares to other methods:
.

  • Parsec Host: In Parsec settings on the PC, go to the Host tab and ensure "Microphone" is set to Enabled (Persistent). This ensures the virtual mic stays active even if the connection drops.
  • Phone 1: Install the Parsec app and grant it "Record Audio" permissions.
    Phase 2: Establishing the Connection
  • Connect Phone 1 to PC: Open Parsec on Phone 1 and connect to the Home PC.
  • Activate Mic Passthrough: In the Parsec overlay on Phone 1, ensure the Microphone icon is toggled ON.
  • Phone 2 Relay: Use a screen-sharing tool (like Discord, Zoom, or a native "Smart View" relay) to stream Phone 1's screen to Phone 2. Note: Since Phone 2 is only for visual monitoring, this minimizes network strain on the PC.
    Phase 3: Accessing the Configuration Interface
  • On the PC (via the Phone 1 remote view), open Buzz.
  • Navigate to the Live Recording tab.
  • Input Selection: In the "Microphone" dropdown, select "Parsec Virtual Audio." This is the crucial link—it tells Buzz to "listen" to the audio coming from your phone in the lecture hall.
    Phase 4: Applying and Verifying
  • Set the Model to Medium, Task to Transcribe, and ensure Faster-Whisper is the selected backend for VRAM efficiency.
  • Click Record.
#

PC are connected to School wifi and The connection internet isnt good so im afraid of Network Bandwidth: For live transcription, latency is less critical than stability. Phone 1 requires a strong upload speed at the lecture hall (at least 5-10 Mbps) to ensure the audio stream doesn't "jitter," which can cause Whisper to hallucinate or miss words.

graceful scaffold
#

anyone got a prompt to feed ANY ai model to make it respond like a gen z/teen from today?

supple trail
hallow thistle
hallow thistle
#

I've seen your message in chat, and well that's awkward.

uneven tinsel
#

if I need to start development with AI to build apps and websites what the best platform can I use it first

hallow thistle
hallow thistle
uneven tinsel
#

?

hallow thistle
uneven tinsel
hallow thistle
#

For simple code or big fix, I usually use Notepad. I'm not really the type of a person to go full on coding AI so.

supple trail
#

made my own trainer cuz i didnt like applios trainer lol

image.png
hallow thistle
#

For more experimental features not found in Applio RVC, there's Codename's RVC fork. BocchiThumbsUp

supple trail
#

and i train on kaggle.

hallow thistle
#

I'm not ready to do the same as you, I still prefer the more-friendly one.

supple trail
#

what

#

idk whats more friendly its just a gui difference ruru_confused

image.png
hallow thistle
#

Don't ask me with just word "what", I don't know what to answer.

supple trail
hardy yew
hallow thistle
#

What am I supposed to say?

hardy yew
#

I asked a normal question, lol

#

These are two completely different apps

hallow thistle
#

You made me stress myself trying to say something for real. Anyways, sometimes I use Microsoft Visual Studio for testing some Python environments. Python is a programming language that itself also has its executable (python.exe). I never have Visual Studio Code installed on my laptop.

hardy yew
#

"stress"? No need to take everything personal, this is a place for discussions after all and I don't mean anything bad

#

I thought you actually meant Code because it's q very common, free and lightweight tool for the job.
Python development in visual studio itself seems uncommon to me. Not even saying that it's bad or good, just wouldn't be my go-to personally

storm pollen
#

.

finite wind
#

I always wondered what separating audios with 0.3 to 0.8 seconds of void at the beginning and the end of the audio clips actually does

hardy yew
#

i didn't think there's any reason to include leading/trailing silence

low shard
low shard
finite wind
# hardy yew WDYM?

The person who gave me this tip said and I quote "If noise remains at the beginning or end of a sentence, or if the waveform is repeatedly sliced during processing, the model will attempt to generate audio whenever it encounters a "silent space" at these boundaries. Simply put, it does not tolerate "emptiness." In other words, unless the model is trained on patterns where the beginning and end of a sentence are consistently silent, it will produce a higher frequency of what are commonly known as "silence artifacts.""

#

I was not sure if it actually did what he said so I wanted to share the question

hardy yew
#

sounds weird to me, that's kinda what applio's silence injection is for

#

it adds some samples of silence so that the model learns it

finite wind
#

that same person is also a manual labor freak so he taught me to do that manually I suppose

hardy yew
#

i mean, things like this are so simple that I think scripting it would be the way to go anyway

#

why do something simple as that manually when it can be just automated at preprocessing step

#

but as mentioned before, i don't think it's needed anyway, considering silent samples are already added at extraction step

finite wind
#

I think those who want to squeeze every ounce of quality from the dataset would do that

hardy yew
#

there's no quality difference between adding silence manually and adding it via code, to be fair

#

it's just appending zeros from one or the other side

finite wind
#

ah I didn't count in the fact that applio automatically puts silences in audio clips when processing

hardy yew
#

yeah, checked just in case, it is there

finite wind
low shard
#

hello, rvc means Retrieval-based-Voice-Conversion, not realtime voice changer

can you elaborate:

  • your pc gpu
  • your pc os
  • the tutorial link you're using
finite wind
#

Oh just one more thing, I'm clipping audios manually and clip durations are not the same

#

should I be worried about that? because there are clips from 2 to 10 seconds

still remnant
#

كيف الحال

hallow thistle
still remnant
#

Basim

hallow thistle
#

Is there anything you would like to get help with?

hardy yew
finite wind
#

let's say I don't use applio's auto slicing

#

and I have variety of clips from 2 to 10 seconds of durations I processed

#

would that be a no no?

hardy yew
#

in that case, good question...

Short answer: not sure and I would love to hear opinion here from someone with better experience/deeper understanding.

Long answer:
from what I can see, during training, samples are fed into the networks in equal-sized segments anyway. That's 0.3-0.4s I think, depending on sample rate.
BUT, each sample is considered one step anyway, so the influence of a longer sample will be quite heavierr than of a shorter sample.
So I would expect that similar-but-not-equal sized clips would be fine, but when there's a 5x difference (like 2s and 10s), the longer ones would cause heavier weight adjustment.

In short, I think equal-sized samples should work better. The generalization should be better and gradient more stable, because each sample should be "equally important" to the model

#

But that's just my understanding based on how it is utilized during training. I didn't ever verify the actual effect, so it's just my theory for now

finite wind
#

shucks

#

I was focusing on not cutting off the sentence structure mid way til 10 seconds(since it was my personal limit for a clip to have) and it's varying from 2 to 10 seconds now :/

#

eh I'll try feed about 16 to 30 minutes of these and let people know on here

hardy yew
#

For the record, my first couple trainings were done in a scenario like yours, where sample sizes were like 2-8s with some outliers being as long as 14s. I think I didn't use the auto-slicing then. The models were quite nice anyway.
But then, it was the beginning of my adventure with RVC so it might have just been "fine to a rookie me", and in reality perhaps they weren't so good. Dunno, i'd have to dig them out from my HDD

finite wind
#

14s is super long damn

#

I'm surprised it was decent quality after all

minor isle
#

do i have to update okada

hardy yew
#

@low shard free money glitch here

finite wind
#

I'm going for that invididually processed multiple audio files so I wouldn't know about breaks between audio files

#

unless I mistaken what you've said

hallow thistle
finite wind
#

like, I already have a hundred ish wav files(2s to 10s short clips) in a dataset folder because uh

#

nobody actually said anything about "you should combine them all together with breaks"

tame glen
#

hello people, are there any working colabs or online spaces to use rvc with or w/o ui?

hardy yew
minor isle
finite wind
minor isle
#

im still using the fork but forgot its name

hallow thistle
hallow thistle
finite wind
#

I'll update once I finish processing this dataset and train it

tame glen
minor isle
#

in games sometimes too

hardy yew
#

But yeah, I'm learning too

#

Love to suck in as much knowledge as I can, but there's still lots of missing pieces for sure xD

neat lagoon
#

guys im stuck

#

plz help

low shard
neat lagoon
#

thx

tame glen
#

does rvc work with comfyui?

#

I am running a workflow but stuck on the first step

hallow thistle
low shard
tame glen
tame glen
low shard
#

what's your pc gpu and os? have you tried checking if you're good to run it locally first tho?

tame glen
low shard
tame glen
low shard
low shard
minor isle
raven burrow
#

How to train model?

hallow thistle
raven burrow
#

Rtx 3060

hallow thistle
raven burrow
hallow thistle
finite wind
#

hey just asking, what do you guys usually do when within a clip, there's a noticeable pause to the sentence from 0.2 to 0.5 seconds?

#

like, for example, "sorry, I mean...(pauses for 0.4sec) I mean, it's nothing shocking, I'd imagine."

#

what do you usually do for that paused part?

#

do you void it out? or do you just let that part be as it is?

#

even some music clips I do atm have pauses in between lyrics

low shard
patent trellisBOT
# low shard -realtime
🔊 Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

minor isle
low shard
minor isle
low shard
minor isle
#

better*

hardy yew
#

I guess it can be done in various ways. Though probably automatic silence slicing from applio won't do a great job here

finite wind
low shard
minor isle
low shard
#

vonovox might have more recent updates

bronze stirrup
#

how do i use voices??

winged adder
#

How to replace person in a photo ?

#

Do you know a tool that let me mask
"Replace masked area with [describe person from file:1, e.g., man with short hair, smiling, same lighting and pose as original, muscular build, white tank top, party background, high detail, seamless blend

hallow thistle
wicked bane
#

hey i need help with installing the voice changer on linux? (gentoo) i have a 3060 gpu

low shard
#

i'm guessing linux gentoo

wicked bane
#

its fine

low shard
wicked bane
wicked bane
viral mason
#

If you're a guy and use a guy voice pretty much always works well at 0

wicked bane
#

or do i need to open a forum

viral mason
#

But it does help opening a forum

left steppe
#

guys

#

is there a unrestricted ai on archlinux

bitter notch
#

nvm got it fixed

strange juniper
#

i found a model i really like but the quality sounds like a wakie talkie and i have no idea what to do about it. ive tried just about everything

cyan kettle
#

hello

#

Can you give me the feature of sending a picture? I want to send a specific picture of settings to help me with it?

#

Who best voice changer
And why?
1- Tg Develop's W Okada Fork
2- Vonovox
3- Applio Realtime
4- Deiteris' W Okada Fork

viral mason
austere whale
#

<@&1159293204038955078>

#

Do i install the latest version of vonovox (beta)

viral mason
austere whale
viral mason
#

here's the dl

#

you have a virtual cable btw right?

austere whale
viral mason
#

it's not though, most likely was the model you used

viral mason
#

haven't tried it tbh, I dunno how good it is

austere whale
cyan kettle
viral mason
#

I can't send the downloads ATM since I'm in the bathroom on my phone

cyan kettle
viral mason
#

-rt

patent trellisBOT
# viral mason -rt
🔊 Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

viral mason
#

extract the first one then drag and drop the second one into the folder of the first one

#

then run mmvcserversio

viral mason
#

I'm heading to sleep now so if you need help ping the helper role

low shard
low shard
low shard
lusty pilot
#

Any pretrain suggestions for singers?
The ones I use tend to blend falsetto and chest voice

low shard
#

Are you trying to do e girl trolling / catfishing like the youtube tutorial?

#

Then why are you using a youtube tutorial for e girl trolling? That tutorial is outdated asf, you should delete everything

noble pendant
#

Hello, can you help me? I once came here to download a voice program, but in the end I couldn’t.

low shard
noble pendant
low shard
#

you shouldn’t reinstall the same thing, you should try wokada tg develop fork

#

-realtime

patent trellisBOT
# low shard -realtime
🔊 Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

noble pendant
#

He said you can basically change your voice to any voice.

low shard
noble pendant
#

And I don’t have any tutorial links either

low shard
noble pendant
#

I just want to make a Zelenskyy voice.

noble pendant
#

and prank my friends

noble pendant
plucky crown
#

Hi ill buy a 3090ti fe hpu 5950x cpu and 32gb ram 3600mhz (might upgrade for more ram soon) pc , am i screwed, will the ai not work/choppy/not real time??

low shard
patent trellisBOT
# low shard -realtime
🔊 Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

low shard
#

try vonovoc or wokada tg develop

low shard
plucky crown
low shard
plucky crown
#

Cover

hallow thistle
plucky crown
#

So is it enough

low shard
# plucky crown Cover

yeah it’s way more than enough, but covers aren’t realtime so it seems you’re confused

plucky crown
#

But how long to wait til the voice comes out

hallow thistle
plucky crown
plucky crown
hallow thistle
hardy yew
#

anyway 3090ti will be more than fine

plucky crown
hardy yew
#

i meant VRAM

#

but 3090ti has plenty

#

nothing to worry about

plucky crown
#

Any experience during training

hardy yew
#

I don't understand the question

hallow thistle
plucky crown
#

The voice training they say it takes multiple days to finish one training

hardy yew
#

It all depends

#

But no, generally no

#

I don't know the exact training performance of a 3090ti but the speed should be decent

#

but it will depend on some factors

#

mostly the amount of data you're gonna train with

hallow thistle
#

What do you think about how I explain?

plucky crown
hallow thistle
plucky crown
#

Just asking what you staffs here use

hallow thistle
quaint apex
#

-realtime

patent trellisBOT
# quaint apex -realtime
🔊 Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

hallow thistle
willow crater
hallow thistle
willow crater
hallow thistle
willow crater
willow crater
hallow thistle
willow crater
willow crater
low shard
left steppe
#

guys

#

i need yalls help

hallow thistle
left steppe
#

what is the most unsensored llm to run locally on low specs

#

laptop

#

on lm studio

hallow thistle
left steppe
#

everything

#

pleaseeeeeeeeeeeeeeeeeeeeeeeeee

left steppe
hallow thistle
# left steppe i ned ittt

I know more about RVC and the voice changer related topics, so sorry if I just came here to ask without waiting for more context. sad_catMika_sad

left steppe
#

ok

hallow thistle
# left steppe ok

What is your laptop GPU? You can try search for "an LLM model that can run either CPU locally" if this might help. Let know here if you have problem trying to install, I might be able to pinpoint an issue.

left steppe
#

im running it on cpu

#

ik its bad but i5 1135g7

hallow thistle
left steppe
#

8

hallow thistle
#

8GB is at least usable, although 16GB or more is more ideal.

viral mason
#

I see so many people wanting to use real-time on a crappy laptop, I feel bad for anyone with intel

tame oracle
#

Good thing I have a still useless Vega 7 then

winged adder
#

How to download Huggins face model

low shard
# winged adder

that's not an RVC Model btw
that's whisper, an STT model, you can use the download icon to download specific files

winged adder
#

I downaloded already and It work

low shard
winged adder
#

Yep

swift thunder
#

¿Where can I find NSF-Hifigan?

nocturne mural
swift thunder
brittle wing
#

Hi, i want to train my own model. Quick question. Is it okay to throw a 2+ hours of data in single file in Applio or should i cut it by parts?

nocturne mural
oblique needle
#

Hey guys, I would like to ask if anybody knows an AI tool that can create avatars, for free
just like this banned account "rabbigoldman" was doing (he is still on youtube and I'm not trying to advertise that channel, I just want to know an AI tool that can mimic the job for free)

swift thunder
viral mason
#

I just got out of the shower and I feel gross again

brittle wing
#

Hi, i want to train my own model. Quick question. Is it okay to throw a 2+ hours of data in single file in Applio or should i cut it by parts?

#

Anyone?

covert panther
#

Hi which real time voice changer is the best for now? I have macOS

viral mason
#

what is your gpu (Nvidia or AMD) and what do u plan on doing with it? just curious

hallow thistle
flat junco
#

cable input

#

i cawnt hear anything

hallow thistle
# flat junco cable input

Don't try to ignore the Sapphire bot here, the bot gives you how to asp properly. What is your PC GPU? Did you follow any tutorial or guide before? And what will you use the voice changer for?

#

Don't skip steps. What is your PC GPU? And what do you use the voice changer for?

hallow thistle
teal arrow
#

i have been messing around with running an ai locally on my own computer
i set up lm studio and loaded google gemma 3 4b and then connected it to a small python script so i can chat with it using the openai style api but everything runs locally
nothing goes to openai or any cloud servers it is just localhost and my own machine. mostly doing it to learn how this stuff works and honestly it is pretty fun
does anybody know how to make my
ai uncensored? and what more could i do to upgrade it
m gonna be trying dolphin‑2.9‑llama3‑8b‑gguf
ram expensive

flat junco
flat junco
#

gpu

hallow thistle
flat junco
#

i did it right but now

#

i wannnnna out in aa voice

#

but idk how

hallow thistle
#

You didn't directly answer my questions.

#

This won't be a time to goof around. To check your PC GPU on Windows, open Task Manager, go to Performance tab, spot if there any GPU 0 or GPU 1, one of them could be a dedicated GPU. If you mean something else, let me know at least.

flat junco
#

gpu 1 ig

#

idk

hallow thistle
#

Why do I have to tell you every step like that? Click on GPU 1 to real its full name on the right panel.

viral mason
flat junco
#

amd

viral mason
#

-rt

patent trellisBOT
# viral mason -rt
🔊 Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

viral mason
flat junco
#

is it safe??

viral mason
nocturne mural
flat junco
#

ok

hallow thistle
#

It would be better to ask for more detail, I know this approach gonna be slower; otherwise we're gonna let the person to use the voice changer with an integrated GPU, which would be quite annoying to get along. nso_ame_shrug

bronze maple
#

Hello, I have a serious problem. I used to train a voice that I was able to use with Applio, but now with the new Applio, when I insert the .pth file, nothing happens. What is the solution?

screenshot_1775696877497.jpg Capture_decran_2026-04-09_030852.png
flat junco
#

@viral mason

#

ccccan u join call 1

#

to help me

hallow thistle
flat junco
#

-_-

#

meanie

hallow thistle
#

Stay here, please.

bronze maple
flat junco
#

or can u support call

hallow thistle
flat junco
#

noo am talking to

#

your local

#

worm

hallow thistle
bronze maple
hallow thistle
#

Also, you can send your screenshot to here now, I'm more of this way.

hallow thistle
bronze maple
hallow thistle
#

Oh no. Well, I actually don't know about how to combine D and G files into one, sorry.

viral mason
#

tried sending in dms since you asked there but I couldn't

hallow thistle
viral mason
#

?

bronze maple
viral mason
hallow thistle
#

I could've go faster, but the bud seemed to made things slower to progress. Not only he did leave the server for whatever reason.

bronze maple
viral mason
hallow thistle
nocturne mural
bronze maple
hallow thistle
#

To get a single .pth file, there are possible ways. Either train the new one from start or use a script that could convert into a single usable model.

#

Using a script would be a faster approach. But did you remember which sample rate was your model trained?

hallow thistle
limber fiber
#

what's the best site to use ai models on? And how do I use a model I've downloaded?

bronze maple
viral mason
#

what kind of voice trolling? like darth vader or goku or smth

hallow thistle
#

What does this mean?

hallow thistle
hallow thistle
bronze maple
hallow thistle
bronze maple
#

wtf 😂

#

Okay, I'm going to redo my model. I have 4 minutes of dataset. What parameters should I use to get the best quality?

#

Because I just did 100 pockets, and it stinks, compared to my model from two years ago, which is excellent

hallow thistle
limber fiber
hallow thistle
limber fiber
#

microsoft 10

viral mason
#

please reevaluate your life

hallow thistle
#

-rvc

patent trellisBOT
limber fiber
limber fiber
low shard
#

elaborate:

  • your pc gpu
  • your pc os
  • what are you trying to do: TTS, Ai Covers, E Girl Trolling / Catfishing, Roleplay
  • the tutorial link you’re using
hallow thistle
#

I checked your older messages, and well that's awkward.

hardy yew
#

If you have an Nvidia GPU, then yeah, get the 17_11 vonovox beta and it should be better than the w-okada forks

#

you can check here for download

#

nah, there's just a bug with using index files in that release and that command fixes it. So if you're not gonna use index at all, you can skip it

#

TBH don't expect a massive quality increase, most of the sound quality is a result of the model after all

#

But it should definitely be more stable, and apart from that, there's lots of improvements in vonovox that are not present in w-okada forks

#

Better pitch/feature extraction, improved stitching, volume handling etc.

#

So in those terms it should be better

#

Whether it's audible will depend

covert panther
hardy yew
#

That's a new

#

WDYM by "install"? Offline inference on audio files or realtime?

#

For training or merging models Applio is still the right path

junior iron
#

is there any alternate method to obtaining okada, the hugging face redirect is currently not working

hallow thistle
hallow thistle
junior iron
hallow thistle
junior iron
hallow thistle
junior iron
#

and its rvc?

hallow thistle
#

Like W-Okada, Vonovox uses RVC voice models.

junior iron
#

i can give it a look for sure

hallow thistle
open pelican
#

where is maeko (Ajthefunky) rvc

junior iron
hallow thistle
#

While using Dione Launcher might be easier for most people, the downside is that it's harder to know what's going with terminal when you get errors. Some people would directly run the batch file instead.

hallow thistle
open pelican
#

i want to unban me on weights discord

low shard
low shard
clear oasis
#

I wanted to know how many channels are posting AI Covers of many famous songs by AI, remaking them with different voices? Including a channel that does AI Covers with Ariana Grande's voice that sounds really good!
Until now, I only know Suno, which many say does well, but there's that copyright thing there that nowadays, you can't do anything, unless it's from a very unknown artist.

terse nexus
#

I wanted to make a website using ai but the ai's then to limit messages. I used claude and it gave me a good code, but there were some errors and things I didn't like but I couldn't make the changes since the messages keep getting limited, and this is taking alot of time. So what should I do now?

cedar rock
#

any advise on how to start fress....a newbie here 😁

hallow thistle
#

What do you use the voice changer for? And are there even two different AMD Ryzen CPU models in the same PC system?

torpid oriole
hallow thistle
#

Which voice model would you like to use?

hallow thistle
viral mason
hallow thistle
viral mason
#

This is super outdated, since you have a 5080 you should use Vonovox

#

2 months old? Anything on yt about Wokada is outdated as they don't use the newest stuff like Vonovox or Wokada tg fork

#

Btw what do you plan on using this for, just curious ^^

#

That virtual audio cable can cause issues on windows but if it's working fine for you you don't have to switch, it's just recommended over VB cable (the one you have now)

#

Alrighty

#

I'm happy to see someone using the voice changer normally and not what most of the screenshots in hall of fame showcase

brave quartz
#

Hi I wanna help where I can find good text to speech ai same as applio ?

hardy yew
viral mason
brave quartz
#

It's not working with me

#

When I use model it doesn't show up the voice or something

brave quartz
viral mason
brave quartz
#

I wonder if I'm the only one who have this problem

viral mason
finite wind
#

hey call me crazy, but I already processed all my audio clips into these and I want to ask

#

should I separate each individual sentence as an audio file because I can do that, or should I just put this single long audio file, like the screenshot I took, into the Applio?

viral mason
#

I personally just use a singular audio file since applio already cuts it up automatically into 30s clips

#

Saves room on my pc

finite wind
#

okay I agree to that

#

but hear me out

#

I already spaced out like the screenshot above

#

and should I still let Applio to cut it automatically?

#

and I want to ask is applio cutting into 30s clips is by default and can be adjusted or

viral mason
#

Ping the helper role to see if anyone like Namari would know, they're pretty smart

#

I'm just kinda someone who knows how to use Applio and the voice changer stuff as well

#

Not super advanced knowledge:p

brave quartz
viral mason
#

I don't personally use the tts function often but you just use any model and type something on the tts section and it should sound like the model you're using

finite wind
finite wind
craggy brook
#

Why won't Applio open? I'm going crazy!

chilly furnace
#

if it's already cut you don't have to cut it

chilly furnace
#

though the default is enough

#

i forgot what it was

toxic perch
#

wich program and model are u using?

#

can i see a screen maybe?

#

i also have amd so i'll try

#

ok so lower the extra all the way down

#

and put index to 0.6

#

then put chunk to the max

#

also the tresh try to put 0.6

#

yeah but for now try with the longest then adjust urself

#

and keep the one that u prefer

#

i mean near the middle

#

a little to the right

viral mason
#

this is so old

toxic perch
#

yep that'a another problem

#

u should use applio but what i see

viral mason
#

I got u 1 sec

toxic perch
#

are you trynna scam?

#

that's mae a girl voice

viral mason
#

it's mainly a voice training software but has realtime function

#

for what reason..

toxic perch
#

@round fog idk if this is catfish

viral mason
toxic perch
#

we are just making sure

viral mason
#

voice models like the one you are using here are commonly used for catfishing/scamming

toxic perch
#

what tutorial have you watched?

toxic perch
viral mason
#

ewwww

#

"ai girl voice changer" that is not what it's for 😭

#

I hate any yt videos like that

toxic perch
#

bro

viral mason
#

just use a normal voice like a game or show character

#

don't use egirl things they're gross

toxic perch
#

ai egirl voice changer

#

that is not it

viral mason
#

like those nasty whisper asmr things

toxic perch
#

@cosmic spire @dusty rampart can you guys check please?

viral mason
#

and they have those weird pictures too

#

that video is very outdated

#

that's why what u have is also old

#

anything on yt about realtime voice changers are outdated since they don't keep up with current stuff

toxic perch
#

we are against

#

the catfishing

#

so if we see even the smallest thing we have to make sure

#

it's not hate against you

viral mason
#

thank you for being normal ❤️

toxic perch
#

and also before anything there are some guides

#

applio is a similar program but updated

#

and it has more functionalities

#

sorry for my englis

#

italian here

viral mason
#

for AMD I recommend Wokada tg fork tho Applio is a option I'm just unfamiliar with it voice changer wise

toxic perch
#

and also

proven hill
toxic perch
#

you can update the drivers, if you have amd adrenaline

#

then also asio4all it's not bad

toxic perch
#

che strano un italiana qua

proven hill
toxic perch
#

hahahaha veramente, comunque se vuoi parlare andiamo su chat che qua si fa assistenza

proven hill
#

oki ma parliamo in inglese altrimenti ci cazziano

toxic perch
#

cause i have some latency issues

#

nah

viral mason
#

nono, since you have AMD you have Wokada TG and Applio's realtime

toxic perch
#

search the latest version online then install it, then in the window chose only update drivers and ur good

viral mason
#

got u, I'll get the link fr this time

#

-rt

#

uhhh

#

hello?

#

-realtime

#

bruh

toxic perch
viral mason
#

stupid bot broke

patent trellisBOT
# viral mason -rt
🔊 Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

patent trellisBOT
# viral mason -realtime
🔊 Realtime RVC

Guides for Programs that use RVC Models in Realtime for Calls/Games

• Vonovox

A Realtime Voice Changer with similar performance to Wokada Tg-Develop Fork, with extra features, but it supports only Nvidia GPUs on Windows 10/11 unlike other options that have wider support. and without cloud options

• Wokada Tg-Develop Fork

A personal fork (modified version) of Wokada Deiteris Fork, it just adds some Quality of Life improvements to it like supporting Spin Embedder and Audio Effects. Don't expect too much about it since the creator made it originally as a personal project.

• Applio Realtime

A Realtime Voice Changer with similar performance to Vonovox & Wokada Tg-Develop Fork, with extra features.

• Wokada Deiteris Fork

Deiteris' fork (modified version) of wokada that doesn't get updates anymore.

⛔ Outdated/Discouraged

These options are not recommended for use.

• Original Wokada

Not suggested, older versions in youtube tuts are even way worse. GUIDE

• RVC GUI Mainline Realtime

The program is worse compared to the ones above, and much less updated. GUIDE

toxic perch
#

there it is

viral mason
#

just the developer name of who made that version of wokada

viral mason
toxic perch
viral mason
#

first is the voice changer second is a virtual audio cable to connect it to games and discord

toxic perch
#

the voice model that u use does a LOT of the work to be fluid

viral mason
#

I try

#

it's very setting dependent and software as well

toxic perch
viral mason
#

of course!

#

I already accepted

#

we have talked enough to where I trust you

toxic perch
#

thank you so much

#

my d&d sessions are gonna be craaaazy

#

if this works out

viral mason
#

for amd ONNX is needed yea

#

voice models aren't for a specific type of gpu

#

they work on all

toxic perch
#

it's what works the best

#

for yours

viral mason
#

weird code probably

#

maybe wrong version

proven hill
#

i need 3 latina, 5 egirl, 2 mommy, 1 asmr, 6 indian and 3 french girl voices and i want them NOW

viral mason
toxic perch
proven hill
toxic perch
#

or just old

#

frfr

#

there is not a newer version

viral mason
#

just delete the old one so it doesn't interfere, the new program is completely separate from the one u had

toxic perch
#

there are new programs tho those ones

#

we trust u

#

dont mess up

viral mason
#

you may add me

#

my dms are open for the both of you to ask or just chat

toxic perch
#

yeah but just started too

viral mason
#

use your new power correctly and safely ^^

#

for me, years

toxic perch
#

like a week

#

i'm a bit o a nerd tbh

viral mason
#

use it if you wanna add cool voice effects to wokada, but the ai voices on voicemod are doo doo

toxic perch
#

oh hell naaawww

viral mason
#

if you buy it, do the lifetime one and then just keep it for the soundboard lol

hexed ruin
#

One message removed from a suspended account.

toxic perch
#

tbh my boyfriend is the one who knows all and he set me up a lot

viral mason
#

paid gives infinite space tho lol

#

spamming it is for losers, use it for comedy and perfectly timing jokes

viral mason
toxic perch
#

l friends

low shard
low shard
terse nexus
#

I tried running it locally but I think it isn't as effective, and I don't know how I can upload file/image in it then

low shard
#

elaborate:

  • your pc os
  • what are you trying to do: TTS, AI Covers, E Girl Trolling / Catfishing or Roleplay
  • the tutorial link
low shard
low shard
#

also please don't ping admins directly, ping mods

low shard
low shard
toxic perch
#

che bello italianiii

low shard
#

there were a bit more before the 2023 rvc ai cover hype died

toxic perch
#

oh i see

#

good to see some anyway

viral mason
#

most are over a year ago tho

gilded robin
#

hey does anyone know something like Zorq AI's motion control but free or client-side? such as facefusion?, but id like something that doesnt replace only the face

glass smelt
#

is there any way to use with intel arc? i have Intel(R) Arc(TM) B580 Graphics

finite wind
hallow thistle
dire spire
#

RVC.