🧬│ai-chat

[covert lake]

if you really want to use it locally, u can https://github.com/rupeshs/fastsdcpu/

but its not the best

if u want

[oblique dragon]

Alright

I just wanna create images

[covert lake]

else would be better to use cloud, the best model rn is flux.1-dev https://huggingface.co/spaces/Nick088/FLUX.1-dev

[covert lake]

u can also use dalle3 (from chatgpt+) for free via @elder willow in #🤖│bots

ofc the best quality one is flux.1-dev

[final void]

hello

[dull saffron]

Is anyone here voice actors

Smiles”

[glass dust]

hi

how can i copy the link of this ?

@covert lake

[covert lake]

you can't put it directly into cloud

what are you using rn?

[glass dust]

like what , pc?

[covert lake]

are you using ilaria rvc on google colab?

[glass dust]

yes

[covert lake]

like ilaria rvc mainline or smt

bc that's broken

[glass dust]

mainline

[covert lake]

yep broken

#📰│dev-updates message

[glass dust]

bruh

[covert lake]

Use ilaria rvc zero which is better

[stray merlin]

Ilaria RVC: CLICK HERE 🤗

Guide on how to use it: CLICK HERE 📝

Don't forget to thank Ilaria if you find it useful! 💖

[glass dust]

thank you bro

[covert lake]

its also written in the guide how to use weights.gg models, read it up

[covert lake]

your welcome

[glass dust]

btw

whats wrong here

it wont let me upload the model

[covert lake]

you can't download directly weights.gg models like that bc weights.gg requires an acc

check this part of the guide pls https://docs.google.com/document/d/1YbXcLFPaGjhOdG5NFkK3QrucCEpHZBwFUxkeMO8aB18/edit?tab=t.0#heading=h.c5ea25jd74zp

[glass dust]

okay

[fathom patio]

How do you make a ai song sound real

[tacit falcon]

Skill

[sand maple]

hi

[fathom patio]

Hmmm

[grand relic]

Did I miss anything?

[feral halo]

Ayo? @grand relic level 1 !!!

[tepid herald]

hi

where find models

[bitter lantern]

come voice channel

[grand relic]

What kind of model?

[bitter lantern]

lets talk with me

[little nebula]

#1175430844685484042

[rough ocean]

hello chat

[grand relic]

Hey

[solar merlin]

@atomic vector settings

[atomic vector]

what do you need help with settings for
-# AI-generated responses may be inaccurate; please verify important information.

[solar merlin]

@atomic vector help

[atomic vector]

Help

Applio is a bot with features like moderation, fun, utility, and more.

AI Hispano

• /model-maker: AI Hispano » Starts the verification to obtain the 'Model Maker' role.
• /model: AI Hispano » Start the verification of your model.
• /punishment: AI Hispano » Apply a sanction to a user (Moderators only).

Info

• /chat: Info » Chat with Applio.
• /avatar: Info » Get the avatar of a user or yourself.
• /help: Info » Get help with the bot commands.
• /invite: Info » Get the invitation to have the Applio bot on your server.
• /serverinfo: Info » View information about the current server.
• /stats: Info » Get statistics about Applio.
• /support: Info » Join the Applio support server.
• /userinfo: Info » Get info about a user or yourself.

Moderation

• /ban: Moderation » Ban a user from the server (Moderators only).
• /kick: Moderation » Kick a user from the server (Moderators only).

Utility

• /afk: Utility » Set yourself as AFK on the server and prevent anyone from mentioning you.
• /ping: Utility » Get advanced information about Discord and Bot latency.

Requested by avatarthewayofwater09 · Oct 21, 2024, 5:13 PM

[covert lake]

its a bot

[tacit falcon]

@atomic vector hello bro

[atomic vector]

hey what's up
-# AI-generated responses may be inaccurate; please verify important information.

[covert lake]

it doesn't know anything about rvc, its just a simple chatbot

yea but still not trained on rvc

for help ask in help channels

[solar merlin]

alright i found it

[feral halo]

Ayo? @solar merlin level 3 !!!

[stray merlin]

Voicechanger Settings (Okada)

⠀

Settings for Nvidia GPUs :nvidiagpu:

F0 Det.: rmvpe (suggested for all series)

RTX 40-series: 80-96 chunk | +16384 extra
RTX 30-series: 96-112 chunk | +16384 extra
RTX 20-series: 112-128 chunk | +16384 extra
GTX 16-series: 128-192 chunk | +8192 extra
GTX 10-series: 128-192 chunk | +8192 extra

Advanced Settings

Protocol : Sio or Rest
Crossfade: 4096 start 0.2 end 0.8
Trancate: 300
Silencefront: Off
Protect: 0.5
RVC Quality: Low

⠀

Made with love by Ilaria 💖

[covert lake]

You can search rvc ai voice models at:

• #1175430844685484042
• Send " @atomic vector search (name of the model)", without the ()
• Do /find with @hidden grotto
• https://weights.gg/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://applio.org/models
• https://voice-models.com/
• https://thevoicemodels.com/ (for Turkish Models, login required with discord and level 2 on their server)

if there isnt one, you can:

• #1159289738314919936
• #1191429836321849435
• make it yourself with our docs guides https://docs.ai-hub.wtf/essentials/how-to-make-voice-models/

[ancient hull]

What's the best AI that can do TTS? Not just words but laughter

[covert lake]

There are different Text To Speech (TTS) AIs:

GPT So Vits: RVC isn't as good as GPT So Vits for tts, but gpt so vits (few shot tts, which means needs just a lil training for models) can't use rvc models (and viceversa), and its only limited to: english, chinese & japanese, if you wanna check gpt so vits instead, read https://docs.aihub.wtf/tts/gpt-sovits/

Freemium 11labs: A easy way to do TTS is https://elevenlabs.io/, you can't use RVC model on this but its a mostly premium easy way for good quality TTS

FishSpeech: FishSpeech is a 0 shot (no explicit training needed) TTS, if you got a good pc you can use it locally else use their site

With RVC Models:

RVC is natively for Speech To Speech, but forks such as ilaria rvc mainline & applio have built in tts (using Microsoft Edge TTS to make a generated tts audio, which i suggest you to choose a tts model that is the same gender and language of the rvc model you wanna use, and then convert it with rvc)

If you wanna do tts locally with RVC Voice Models (if you got a good pc):

• You can get Applio in our docs
• While Ilaria RVC Mainline here (no guide as of right now)

If you don't got a good pc you can do tts with RVC Voice Models on cloud:

• Ilaria RVC Zero (Running on A100 GPU, free fasted rvc on cloud) and the guide

• Use Applio UI Colab (with google colab T4 free daily limit gpu)

• if you don't wanna use edge tts, you could try another tts ai from our tts index and use the output as an input in rvc

Gpt so vits and 11labs = the best

Not sure if they can do laughs

[light bone]

gpt sovits can if u have it in dataset

but its random af

[covert lake]

Oh lol

[tame solstice]

Hi, where I can get the AI RVC makers?

I just can't access them

I could before but now I can't

What's going on?

[amber light]

wdym by that

“rvc makers”

[tame solstice]

In the info where it says: "Making AI Covers with RVC", I click on the link and it says I don't have access to this adress

[severe turret]

hi where do i donwload voice changer ?

[tame solstice]

Everytime I click on the link I get a window with a lock and it says something like: "You don't have access to this adress. This is a link to a server or channel you do not have access to" And this had never happened to me before

[feral halo]

Ayo? @tame solstice level 3 !!!

[tame solstice]

Do you mean from the info section?

That's what is happening to me right now

I can't access the link where the links are where I can make models with RVC

[light bone]

tf u are even talking about what info section😭

[tame solstice]

#1159949278270193734 this

[amber light]

what gpu do you got

if any

[amber light]

what

-rvc

[rare sorrel]

📚 Documentation

AI HUB Docs

https://docs.ai-hub.wtf

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde
• Italian Guides by Ilaria

[severe turret]

i 5 10 gen

[amber light]

i said gpu

lol

not cpu

[severe turret]

oh mb im in middle of league game

1650

gtx

[amber light]

hmm

it’s possible to run it’s just very heavy on your gpu

[severe turret]

yea ik i used it before

[feral halo]

Ayo? @severe turret level 1 !!!

[amber light]

-realtime

[rare sorrel]

This interaction has expired, use the command /guides if you wish to see it again.

[amber light]

there’s also online options

[stray merlin]

Voicechanger Settings (Okada)

⠀

Settings for Nvidia GPUs :nvidiagpu:

F0 Det.: rmvpe (suggested for all series)

RTX 40-series: 80-96 chunk | +16384 extra
RTX 30-series: 96-112 chunk | +16384 extra
RTX 20-series: 112-128 chunk | +16384 extra
GTX 16-series: 128-192 chunk | +8192 extra
GTX 10-series: 128-192 chunk | +8192 extra

Advanced Settings

Protocol : Sio or Rest
Crossfade: 4096 start 0.2 end 0.8
Trancate: 300
Silencefront: Off
Protect: 0.5
RVC Quality: Low

⠀

Made with love by Ilaria 💖

RVC Colabs and Spaces

⠀

Google Colabs :colab:

⠀
AICoverGen-WebUI
Useful for making quick covers, by Hina.

AICoverGen-NoWebUI
Useful for making covers, doesn't include a UI, by Ardha, by Eddy, Hina and Gdr.

RVC Disconnected
To train new voice models, by Kit Lemonfoot.

EasyGUI
The OG interface, by Rejects.
⠀

Made with love by Ilaria 💖

[spare lotus]

How do people make these 80s dark fantasy movie adaptations?

[real sentinel]

nop

[real sentinel]

Thats a nice gpu bro

[severe turret]

ok so what i was playing game and typing at same time

why u need to make fun of me ?

is your life that sad ?

[real sentinel]

i mean my life is sad

but it doesnt matter here lol

[covert lake]

#1159290752195633273 or else it gets deleted

[placid tinsel]

🤭

[real sentinel]

mod lol

[elder willow]

yooo wassap

[cerulean marlin]

guys what does epochs mean

[feral halo]

Ayo? @cerulean marlin level 1 !!!

[cerulean marlin]

and where do i find it

[covert lake]

https://docs.ai-hub.wtf/rvc/resources/epochs-tensorboard/

[cerulean marlin]

oh its a big process

i thoyught it was just a setting on wokada

[covert lake]

Nope

It's a training info

More doesn't equal better quality if u were already thinking that

[cerulean marlin]

huh

so im downloading a voice model and it tells me to use 95

[covert lake]

It's an info of hoe the RV model was trained

[cerulean marlin]

epochs

[covert lake]

You don't have to worry about it

[covert lake]

Tells u to use 95?

[cerulean marlin]

yes it says like 95 epochs next to the voice link

[covert lake]

That's just the training info, it's not a setting that you can change anywhere

Just don't think about it, dw

[cerulean marlin]

ohh i thought i had to put it at 95 epochs for it to sound more like the model

or smt like that

[covert lake]

It's like when u bake a cake and tell people the ingredients, people can't change those ingredients but it's good to know them

[cerulean marlin]

yea

[cerulean marlin]

do you know what the best setting on wokada is so it sounds better or more realistic?

[ember hound]

im trying to make very convincing ai music, is there any tips that could possibly make vocals sound clearer/more realistic?

[gritty dove]

this is not anime vanguards

[scenic fossil]

lol

[umbral jungle]

new lvl lmao

[half sleet]

wait

[viscid hollow]

the lvlup msg from bot was more interesting that update itself

[half sleet]

this isn't the Crustacean Nation...

[brittle raptor]

u guys will release annything except a new model

[forest heath]

Hi guys.

[sly pecan]

Hi

[junior axle]

W update 🔥

[lethal flax]

LET'S GOOOO FREE TRAINING IS BACK

[dim epoch]

Wait holy crap really?

[tribal horizon]

hey there

[dull elk]

yo

[flat palm]

skibidi sigma

[elder willow]

in ohio, ong?! +10k aura

Dang

[crisp tusk]

LETS GO free training, I had models that I made because I thought it was originally free but then it turns out it wasn't

[feral halo]

Ayo? @crisp tusk level 2 !!!

[crisp tusk]

SHUT UP

[ember hound]

i didnt consider that

highly appreciated

[stoic garnet]

ok ngl i used the song i see the want to in your eye by conway twitty and using elvis preselys voice to sing it worst part is it actually works really well

[tacit falcon]

gone

idk lol i live under rock

[ionic pumice]

lmfao

[broken vapor]

is there any thing that will help me create a yolo detecion ai

[frail sparrow]

[broken vapor]

pog

[lethal flax]

ALSO THEY ADDED "DOWNLOAD TRAINING SOURCE AUDIO" FOR DRAFTS

[main zephyr]

Hi hi, someone here with experience with suno on "Covers"?

[lethal flax]

GUYS JUST A WARNING

CANCELLING A GENERATION DOES NOT REFUND (TRAINING MODELS)

NOT SURE IF THIS IS INTENDED OR NOT

[amber light]

https://media.discordapp.net/attachments/1275188648035483649/1286804725949399133/image0.gif

[amber light]

so real

[lethal flax]

Is this supposed to happen?

[obsidian nymph]

a while back i installed rvc gui off github, and eventually deleted due to lack of use. i recently tried to reinstall, and i was met with multiple error messages even after troubleshooting and installing all dependencies. is the new version broken?

[lethal flax]

I'm guessing not

[amber light]

dont know

[elder willow]

Hey guys, I have tried weight today and I have the screenshots of it, like to see?

[chilly lake]

no, simply nobody bothered to update the requirements, so it all goes sideways. So just download the prebuilt package

[tacit falcon]

likely older version of python required lol

[obsidian nymph]

damn

[tacit falcon]

usually with these projects i have to use a 4 or 5 year old version of python to download the dependencies and additionally to also get it to work

im using 3.8.10 for rvc

[chilly lake]

3.10.11 is fine to use

for pytorch projects with nvidia, usually no higher than 3.12.3-ish

and only if it actually supports 3.12

[tacit falcon]

nvidia 🔥

[loud haven]

hi

[serene panther]

i trained my first model. . .how do i use it or something?

[autumn acorn]

yo

[feral halo]

Ayo? @autumn acorn level 1 !!!

[autumn acorn]

i get 100% cpu can anyone help me

[fringe haven]

What even is training a model

[serene panther]

what do you mean i just trained my own ai voice, you know the update? i did an airhorn

[amber light]

is you training with a cpu

lmao

[autumn acorn]

idk i just followed the guide idk what im doing wrong

do you mind helping?

i got the fork one that opens in ur browser

[amber light]

im assuming you're doing it locally

do you have a gpu

[autumn acorn]

yes

i think its the radeon 7600

[chilly lake]

you can use Applio with Zluda or RVC with DirectML

[amber light]

^^

[chilly lake]

would not recommend directml for training

[autumn acorn]

how do i do that

[chilly lake]

https://docs.applio.org/getting-started/installation#amd-gpu-support-windows

[worthy atlas]

*Hello everyone, I came here to ask for help from those who really want and can help, on these rainy days my computer where my work is, such as specifications, measurements, orders and everything else I need

[autumn acorn]

is this a whole different voice changer?

idk if applio is realtime

[chilly lake]

not realtime

if you need realtime voice changer then you need another app

https://rentry.co/forkvoicechangerguide#download-amd-intel-and-cpu

[open fern]

whats that new ai image generator

[calm raptor]

Guys

Does anyone know the ai where people put characters infront of others

https://youtu.be/IxwrBbaYOgY?si=aDjP-y41Gb2XLjG3
Sort of like this

[gloomy bay]

hi

can someone give me the link for rvc collab

ui

[quartz flame]

[random bridge]

-colab

[rare sorrel]

📒 Google Colab Notebooks

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

More commands: -colab, -uvr, -karafan, -overtrain, /help

[solar torrent]

From the premium training pass turned into high-quality pass now.

[scarlet solar]

Hi

where is the download link for voice app?

[grand relic]

Which app?

[loud jewel]

[feral halo]

Ayo? @loud jewel level 1 !!!

[solar torrent]

-rvc

[rare sorrel]

📚 Documentation

AI HUB Docs

https://docs.ai-hub.wtf

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde
• Italian Guides by Ilaria

[solar torrent]

[midnight spoke]

there is a new model called f5 tts they say it is good. can this model be used for direct speech to speech conversion instead of text to speech?

[paper furnace]

which one should I choose

in the first screen

[zealous zephyr]

hey guys!

[serene kraken]

Does anyone of you knows any free alternative for crayo?

[tired jasper]

anyone training beatriceV2 models?

[glad shard]

Hi what GTX CABLE driver should I download to work with VoiceChanger on my GTX 460 graphics card?

Привет какой драйвер GTX CABLE мне следует загрузить для работы с VoiceChanger на видеокарту GTX 460?

[grand relic]

I've not used that, but tesseract does the job pretty well

[covert lake]

I don't use Wokada, maybe ask in #🔍│help-w-okada

[covert lake]

Nope, use RVC V2 models

[tired jasper]

i'm training something in beatrice, but realized don't worth to create only one speaker for beatrice since the time of training is the same

[covert lake]

I don't really think so, it's TTS

[tired jasper]

i'm planing to create multispeaker beatrice of genshin impact

[covert lake]

Ohh, well most people don't do that as no one knows how to make good Beatrice V2 models

[tired jasper]

i never played this game, but it's a good point to start learning

[covert lake]

Thought u were just someone using Wokada for a sec

Then I remembered u

[tired jasper]

why "good beatrice", people tried and it was bad?

[covert lake]

Afaik, @icy pendant tried but it didn't came out good

Maybe he can explain it to you better

[tired jasper]

oh, ok

ths

thx*

[covert lake]

Flux.1-dev Is the best open source model

[covert lake]

Yw

[tired jasper]

actually flux.1-pro is the best since it's the full model, but pro version is not allowed to use locally

[covert lake]

I should have specified open source

[solar torrent]

Any other version than Stable Diffusion 1.5 isn't recommended to run with only CPU or slower system.

[covert lake]

use fastsdcpu 🔥

[hoary geyser]

hi:3

[mint sage]

7 attempts of use any character's voice model?! 🤔

[tired jasper]

@covert lake
It's in the middle of first step yet, you really think it's bad?

not even in the middle of first step to be honest, it's like 2000/10,000 yet

training just started and the voice is very close already

more clear

maybe the guy that tested beatrice_trainer before used a bad dataset

[grand relic]

Is this going to be a live translation?

[desert barn]

не ну это какащке 52

https://tenor.com/view/dreamybull-ambasing-ambatukam-gif-26992590

https://tenor.com/view/alien-cat-gif-12994496725234862405

[tired jasper]

i'm just showing how training is going, i don't speak mandarin

[grand relic]

My bad

[tired jasper]

the voice can be used as realtime as well

it's a beatriceV2 model

i'm now downloading the Zenyatta dataset to make another training

Zenyatta from Overwatch 2

i love how BeatriceV2 model is much easier to train and gives better results than regular RVC

transcription is not required and i love it

[earnest dragon]

No it has worse quality then rvc but runs faster

[covert lake]

not home rn, cant listen

[tired jasper]

beatrice is much more stable than RVC

[buoyant star]

they updated to v2 beta 2 two days ago. wonder how much it has improved

[tired jasper]

i feel beatrice quality much better

stable and clear

[buoyant star]

good to hear

[tired jasper]

how you can call this worse than RVC?

[buoyant star]

well, v2 beta 0 was trash

[feral halo]

Ayo? @buoyant star level 1 !!!

[buoyant star]

im gonna test beta 2 later

[tired jasper]

i will now train Zenyatta voice in beatriceV2

[buoyant star]

how long did it take to train and what is the dataset length?

[tired jasper]

idk, just experimenting, but most audios was 10 seconds length and all the dataset around 600 samples

up to 10 seconds length*

[dense thistle]

Hlo

I have js joined the server

[tired jasper]

a full training could be very long, but this one i showed before was around 30 minutes

i'm using 3060 12GB

[dense thistle]

I'm looking forward for learning AI technology here

[tired jasper]

learning to use and finetune is easy if you already have python knowledge, the hard part is to create your own AI

[dense thistle]

Ik nothing about python still I'll be able to learn ?

Where I can learn python ?
I js checked the #1159513888199540817
But I don't think python was there

[feral halo]

Ayo? @dense thistle level 1 !!!

[tired jasper]

to use? yes, but to finetune only if people make GUI for it

youtube is good place to learn python

it's not hard actually, just search for "python for beginners"

[dense thistle]

So python is the first thing I need to learn

[tired jasper]

depends

[feral halo]

Ayo? @tired jasper level 9 !!!

[tired jasper]

what AI you plan to use most?

[dense thistle]

Alright I'll start from today

[dense thistle]

For voice specially

Like to create a voice cover for a vdo

[tired jasper]

for voice people mostly use RVC, which have GUI to train

programming knowledge is not required for RVC

[dense thistle]

Can you elaborate what's RVC and GUI ?

[tired jasper]

Retrieval Voice Change (i guess)

GUI is user interface, so you can use it by clicking buttons and filling boxes

https://www.youtube.com/@Jarods_Journey

this guy is an angel, follow him

[dense thistle]

I see and on what platform ? Is there any specific app or we do it on Google?

[tired jasper]

you have GPU?

[dense thistle]

Thank you so much for guiding @tired jasper

[dense thistle]

No

[tired jasper]

the only requirement for work with AI is GPU from Nvidia

but, in this channel i sent he explain how to use BeatriceV2 voice changer model in CPU in realtime

it's a good point to start

most AI requires GPU for fast inference, but a few can run well on CPU

BeatriceV2 (RVC variant) can do it very well

[icy pendant]

Amd works too, just not on original wokada

On fork wokada and rvc realtime it works good

An igpu can run decently on fork too

[tired jasper]

why fork wokada?

wokada seems pretty good for me

[dense thistle]

I see, thn that means I can still start

First I'll make a tight grip over the topics

[polar flax]

have you even tried to train the same dataset in both beatrice and RVC? show the test results if so

[tired jasper]

no, i didn't, but i notice how fast and stable beatrice is in the model they gave to community to test. And the quality of this model is not bad at all

when RVC recognize voice or similar sound it start to hallucinate, beatriceV2 don't do it

even about noise in the voice, RVC hallucinates

[icy pendant]

Lower gpu and cpu usage
Less delay
Higher quality

[polar flax]

• what kind of hallucination? similar to what TTS models may experience?
• how is the amount of artifacts compared to RVC?
• what kind of GAN beatrice uses? (RVC uses hifigan which is inferior to evagan and bigvgan v2)

[chilly lake]

it does not hallucinate, it simply transcribes audio input into tokens using contentvec/hubert and the transcribed tokens are voiced by the model

[feral halo]

Ayo? @chilly lake level 26 !!!

[chilly lake]

yes, hubert does trascribe anything in a human speech vocan range into tokens, so it does produce garbage if you run it on, for example, a song without separating vocals

but it can extract speech mixed with low volume music canceling it

[placid zephyr]

can anyone tell me i have index and pth file how i convert it into a hugging face link

[chilly lake]

put them into .zip, upload .zip to your huggingface

[placid zephyr]

yeah how i uplod imn hufgging face i dont know how to open what open where i put how i put

[placid zephyr]

what if i will give you zip file can you make it in hugging face link and then send me

[chilly lake]

make a huggingface account

[covert lake]

there is no oython guide here

only guides about some ai things and mostly RVC & Wokada

its a server mostly about speech to speech, not much about coding

[tired jasper]

16.flac is the original voice i want to clone (sample)

original_06.wav is the original sample to convert (input)

beatrice_zenyatta.wav is beatriceV2 with 2000 steps of training (output)

how people say this is bad ????

[polar flax]

that has 24k sample rate (compared to 32k, 40k, 48k in RVC) and sounds robotic like in most sci-fi themed movies/games

[tired jasper]

but Zenyatta is a robot

sample rate is not relevant here

24k is not bad too

[polar flax]

try training real human voices so it would be better to compare

[tired jasper]

this is real human voice

[polar flax]

as comparison, mp3/lossy codecs in 128 kbps have 16khz cutout (which means 32k), also humans can still hear in 12-16khz

[tired jasper]

i'm training only up to 4000 steps

the full training is 10,000 steps

[tired jasper]

i'm trying to figure out how to run beatrice as vst3

[covert lake]

[buoyant star]

Instead of using an input file to convert, can you use your mic to say something and record it?

[covert lake]

Are you trying to use it realtime for calls? Or just recording it for smt u wanna say

[buoyant star]

oh sorry. It was supposed to be directed at @tired jasper

[tired jasper]

yes, i can do it with w-okada

but i won't now

[tired jasper]

"in_sample_rate": 16000,
"out_sample_rate": 24000,

it can be changed and means nothing

you can just upscale it later with Apollo if you want too

[polar flax]

• the former has bottom line noise and there's some clicking noise in the "わかりました" part
• the 2k step sounds a bit robotic
• both the dataset and sample to infer have background noise, or did you intend to test the "robustness" against it?

[tired jasper]

maybe let training to finish the steps it can be fixed

[dense thistle]

I see, I'm js curious to learn
That's why I asked questions

Idk why but you sound rude for this

Am I not allowed to have conversation of my interest here ?

[feral halo]

Ayo? @dense thistle level 2 !!!

[polar flax]

python has official tutorial and docs that are quite easy to learn, though there are tons of other python courses you can find

[covert lake]

Sorry I didn’t meant it like that, I meant to say that there aren’t python/coding guides in here as we have guides mostly about RVC & Wokada, but yea u could find guides online like in https://www.w3schools.com/python/ or especially for ai you could check the docs of the biggest ai platform, huggingface: https://huggingface.co/docs (python still)

[dense thistle]

I see, I don't have much knowledge in it so far I'm very new to this

But I'm giving it a thought to start learning out of curiosity

[polar flax]

yea you can learn basics of programming there

[covert lake]

Also, if u are a masochist, try javascript

[dense thistle]

Oh I see, I have misunderstood you thn, I apologise for this

And thank you for clearing the cloud

And this link, is that beginner's friendly?

[covert lake]

well, w3school is pretty good, i used it sometimes for html as they give guides for multiple languages, should be beginner friendly enough

[covert lake]

btw the joke here is that javascript is good but considerated as hard and stupid sometimes https://www.youtube.com/watch?v=J6yiNbiiCzA

[dense thistle]

I see, I joined bcoz I knew I would learn a lot from ppl and server here

[polar flax]

esp the promise, async/await thingy, and bloated performance (though still viable for frontend)

[covert lake]

about the huggingface doc link, well u should start from basic python before doing any AI coding

[polar flax]

that's also my handbook for web programming & DB

[feral halo]

Ayo? @polar flax level 45 !!!

[buoyant star]

The thing is, AI HUB mainly focuses on how to use the program, but not on how to create the program

[covert lake]

yea that’s true

[polar flax]

you can also learn coding when modding/tweaking the programs

[covert lake]

its mostly just on using rvc & wokada, not really about learning how to code general ai things, such as coding a gradio interface for using stable diffusion

[buoyant star]

true

[covert lake]

yea i learn by seeing code too, i think he meant there are no guides about actual coding tho

[covert lake]

yea was thinking of learning also JS (javascript) with that site, i learnt a bit more of html with it already

[dense thistle]

Yeh I'll start from this evening
Btw I found this vdo on yt
https://youtu.be/kqtD5dpn9C8?si=I59juVtLYNRvRMMt

[alpine niche]

#ask What application do you use to generate text to speech?

[tired jasper]

real time beatriceV2

[polar flax]

Zotac: no it's just a 4070 Ti super
https://www.tomshardware.com/pc-components/gpus/zotac-denies-recent-rtx-5090-gpu-boot-up-rumor-vendor-clarifies-that-the-gpu-was-an-rtx-4070-ti-super

[tired jasper]

also realtime

[dapper drum]

https://cdn.discordapp.com/attachments/1296864641690894366/1296878175757799494/image0.gif

[tired jasper]

i'm training a new beatrice model with updated test samples in english and portuguese, some with intentional noise to see what happen

i will let it finish up to 10k steps

it's Mercy from Overwatch 2

[junior grotto]

where do i download w okada?

[mystic junco]

Can I post models made in weights.gg?

[tired jasper]

https://github.com/w-okada/voice-changer/blob/master/README_en.md

[junior grotto]

thanks

[feral halo]

Ayo? @junior grotto level 2 !!!

[covert lake]

goodluck bro

[random bridge]

-colab

[rare sorrel]

📒 Google Colab Notebooks

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• AICoverGen-WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Modified W-Okada's Voice Changer, Google Colab
• 🆕 FaceFusion UI, by Nick088 Google Colab
• 🆕 FaceFusion NO UI, by Nick088 Google Colab
• 🆕 EasyGUI, by Rejekts Google Colab

ℹ️ Note

While the Colab free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

More commands: -colab, -uvr, -karafan, -overtrain, /help

[rose sparrow]

BTW,is it safe to download applio by IA Hispano ,again?
last month it was hacked

[covert lake]

yea its all safe now

what’s ur pc gpu btw?

[rose sparrow]

rtx 2070

[covert lake]

alr good enough to train and inference locally

[gleaming raptor]

Hey everyone! I'm currently working on an app and running into some challenges. One major concern is the potential high costs associated with using external APIs as my user base grows. To tackle this, I want to integrate a local AI solution directly into the app’s files instead of relying on those APIs.

I'm exploring options like TensorFlow Lite or even developing my own AI model. My goal is to find a solution that's as powerful as the latest models but lightweight enough to avoid significant storage issues. Scalability is also a priority, as I want to ensure users can send many requests without overwhelming the system or incurring high costs. Any insights or suggestions would be greatly appreciated!

It'd be great if the AI is in the app files when downloading it

[tired jasper]

trying to figure out how to use beatrice for files instead of stream

[chilly lake]

slice a wav, send chunks to the model, assemble back

[tired jasper]

The problem is to find the code to do it. Also, really need to slice?

[chilly lake]

realtime voice converters usually use small chunks for inference

[tired jasper]

yes, that's what i'm trying to change. using file instead of stream

the buffer will be larger as well

[gleaming raptor]

Does anyone know anything about text generative AI here?

pls dm me

[subtle fossil]

Hello, is tryreplay.io safe to use?

and is it any good?

[hazy quiver]

where can i download

#✨│ai-help

[glad panther]

Where do you download i forgot

[covert lake]

download what, and what's ur pc gpu

[glad panther]

He means rvc

[covert lake]

yea should be the official replay, never used it tho

[covert lake]

u mean rvc?

what's ur pc gpu?

[hazy quiver]

Where can I see GPU?

[glad panther]

620m

Nvidia

[covert lake]

You can check your pc gpu via:
ctrl+shift+esc (task manager) -> Performance tab -> GPU

[covert lake]

yea not good enough

u can't do it locally (on ur pc)

u have to use cloud (remote good pc)

[glad panther]

I used it before i remember

[covert lake]

are u lookin for training (remote good pc)

i think u used cloud before

[glad panther]

But i changed my ssd so now i cant find it

[covert lake]

or cpu which is super slow

[hazy quiver]

AMD Ryzen

[glad panther]

I used cpu

[hazy quiver]

5600G

Radeon Graphics

[covert lake]

thats super slow

your gpu is not good enough to do it locally

you lookin for ai covers right?

[hazy quiver]

me?

[covert lake]

that's a cpu

[hazy quiver]

where can I see GPU?

[glad panther]

Ye

[covert lake]

top right

[covert lake]

use ilaria rvc zero, a zerogpu (A100) huggingface space which is the fastest free cloud

[stray merlin]

Ilaria RVC: CLICK HERE 🤗

Guide on how to use it: CLICK HERE 📝

Don't forget to thank Ilaria if you find it useful! 💖

[hazy quiver]

AMD Radeon(TM)

[covert lake]

that's integrated graphics

meaning its bad

you have to do it cloud too

you lookin for ai covers right?

[hazy quiver]

downlod link?

download*

[glad panther]

Is it just like the normal one?

[covert lake]

none, your pc is too bad to do it locally

[glad panther]

Or is it like worse in a way

[covert lake]

yes quality doesn't change

all rvc versions have the same quality

since a year

[glad panther]

Oh thats nice

[hazy quiver]

what is a other good AI software that's free?

[covert lake]

none for your pc

you have to use cloud

you lookin for making ai covers right?

[hazy quiver]

what's cloud?

[covert lake]

u run the program on a remote good pc, instead of your pc which would be said "running the program locally"

[hazy quiver]

is that like a feature thing?

[covert lake]

Are you looking to make ai covers?

[feral halo]

Ayo? @elder willow level 1 !!!

[covert lake]

its not a feature,
You can run a program directly using ur pc power (locally)
Or run it on a remote good pc (cloud) (its for when u got a bad pc)

[hazy quiver]

does geforce rtx 4060 work?

[covert lake]

@elder willow don't make any stupid wars

[covert lake]

yea that would work for doing it locally but u said u got integrated graphics

[hazy quiver]

well yea but maybe I can get it for my bday or idk.

[covert lake]

just reply if you are looking to make ai covers

i can give you the cloud version

[hazy quiver]

sure

send it in DMS if u want

[covert lake]

are you looking to make ai covers right?

[hazy quiver]

like make?

[covert lake]

do you want to create ai covers, or ai models?

[hazy quiver]

models is like rcv? for voice changer.

[covert lake]

yes, rvc (retrieval-based-voice-conversion)

[hazy quiver]

rvc.

[feral halo]

Ayo? @hazy quiver level 2 !!!

[covert lake]

let me give you all the options:

• Training (make rvc models)
• inference (use models) on pre-recorded audios
• inference on realtime for calls

there are different programs so thats why u need to tell me what u are going to do

[hazy quiver]

how do I do training?

[sonic atlas]

is there any model for the tiktok ai voice

[covert lake]

As you dont got a good PC, its better you use cloud (remote good pc) for training an RVC Voice Model:

• Google Colabs (4 hours of daily gpu for free, not much hours, but easy to use):
  • Applio (ui)
  • Mainline (UI)
  • RVCDISCONNECTED (no ui)
• Kaggles (a bit harder to use and needs phone number but gives 30 hours weekly of better gpus):
  • Mainline (UI)
  • Applio by Vidal (UI)
  • Applio by Shirou (UI, no guide as of right now)
• Lightning.ai (Kinda hard, needs login, no issue with web uis or anything, but only free 15 credits monthly):
  • Mainline (UI, No guide as of right now)
  • Applio (UI, No guide as of right now)

[hazy quiver]

this is confusing

[covert lake]

Google Colab = Ez but low gpu time
Kaggle = SUGGESTED hard but way higher gpu time

[covert lake]

it isn't a one click thing, you have to read the guides

Id suggest you to read https://docs.ai-hub.wtf/rvc/resources/datasets/ and then use kaggle mainline

[covert lake]

then for making ai covers, you can use ilaria rvc zero

[green wedge]

Эййй пацандре есть кто рашенс

[dry scarab]

hey nick

can you help me?

[tired jasper]

omg, beatrice v2 beta 2 as vst3 plugin is like: graal of RVC, MUCH FASTER than w-okada and even more stable

[dry scarab]

I downloaded the audio thing but how do I make it say stuff?

[covert lake]

the audio thing but how do I make it say stuff
What are you talking about?

And what's ur pc gpu?

[dry scarab]

Like I downloaded the ai cover but I want to make it say things

[feral halo]

Ayo? @dry scarab level 1 !!!

[dry scarab]

like say i downloaded a spongebob cover, How can I make it say something like "mister krabs"

[covert lake]

so, you downloaded RVC (Speech To Speech) and wanted to use it as Text To Speech (TTS)?

[dry scarab]

Yes

[covert lake]

what's ur pc gpu first of all?

You can check your pc gpu via:
ctrl+shift+esc (task manager) -> Performance tab -> GPU

[dry scarab]

AMD Radeon RX 6700

[covert lake]

did you download applio with zluda? https://docs.applio.org/getting-started/installation#amd-gpu-support-windows

[dry scarab]

whats that?

[covert lake]

yea that means ur rvc isn't even using your gpu

[dry scarab]

oh wait

nvm I got it

[covert lake]

amd gpus need special things to work bc they aren't much supported, you need to install applio with zluda to use your amd gpu

[dry scarab]

I got it now, thanks for the help

[covert lake]

wait

There are different Text To Speech (TTS) AIs:

GPT So Vits: RVC isn't as good as GPT So Vits for tts, but gpt so vits (few shot tts, which means needs just a lil training for models) can't use rvc models (and viceversa), and its only limited to: english, chinese & japanese, if you wanna check gpt so vits instead, read https://docs.aihub.wtf/tts/gpt-sovits/

Freemium 11labs: A easy way to do TTS is https://elevenlabs.io/, you can't use RVC model on this but its a mostly premium easy way for good quality TTS

FishSpeech: FishSpeech is a 0 shot (no explicit training needed) TTS, if you got a good pc you can use it locally else use their site

With RVC Models:

RVC is natively for Speech To Speech, but forks such as ilaria rvc mainline & applio have built in tts (using Microsoft Edge TTS to make a generated tts audio, which i suggest you to choose a tts model that is the same gender and language of the rvc model you wanna use, and then convert it with rvc)

If you wanna do tts locally with RVC Voice Models (if you got a good pc):

• You can get Applio in our docs
• While Ilaria RVC Mainline here (no guide as of right now)

If you don't got a good pc you can do tts with RVC Voice Models on cloud:

• Ilaria RVC Zero (Running on A100 GPU, free fasted rvc on cloud) and the guide

• Use Applio UI Colab (with google colab T4 free daily limit gpu)

• if you don't wanna use edge tts, you could try another tts ai from our tts index and use the output as an input in rvc

[oak cosmos]

can a 3060 run it well or do i have to use cloud as well

[covert lake]

gpt so vits is better than rvc for tts

rvc is built for sts (speech to speech)

[covert lake]

run what well?

[oak cosmos]

rvc

[covert lake]

yup both training and inference locally

As you got a good PC, you can use RVC locally, you can choose between:

• Applio: A fork of RVC with some extra features like Applio TTS, same quality tho
• Mainline: The original RVC

(do not follow yt tuts, they're old)

[oak cosmos]

which do you recommend?

[covert lake]

they both have same quality, applio might have a simpler ui tho

[tired jasper]

mercy is the only that finished the training

the others i stopped at 4000 steps, so, it's normal to get some artifacts

[solar abyss]

quick question about weights, do premium users get to train other models besides FLUX (I'm asking about SDXL in particular)? thanks

[solar abyss]

what did you train this with? quality is pretty good, mine always sounds like a completely different person than the dataset 💀 no matter what i do

[tired jasper]

it's beatrice v2

i made the inference in real time

maybe you're forgeting to change the transpose

[ancient nymph]

how do i add rcom to path

[feral halo]

Ayo? @ancient nymph level 1 !!!

[ancient nymph]

yeah i figured it out

thank you

but um

how do i add the voice model

i wont let me choose files

[chilly lake]

downoad whan you want, unzip into logs/model_name

[ancient nymph]

thanks

[covert lake]

#1159290752195633273

[icy breach]

Hiiiiii

[slim blaze]

hi

[feral halo]

Ayo? @slim blaze level 2 !!!

[thick hare]

hi

i have prblem i wanted to make an ai cover but when i want to opne RVC-GUI.bat it says: Running with the system Python.
Nie moPress any key to continue . . . but when i press any button it closes i dont know what to do

[real sentinel]

same

[ancient nymph]

i need help i get an error when i do it

when i convert

it says error connection out

connection errored out

rocblas error

[tired jasper]

regretful of download large dataset in single thread

[vestal leaf]

Hellooooo

Hmm...there's no AI music channel

[pseudo eagle]

does anyone know how to make 2 voices sing at the same time? i use audacity as my audio editor.

[amber light]

to avoid copyrighted material being sent here

[covert lake]

if ur lookin to share an ai cover, pubblish it on yt and share the yt link in #1159290752195633273

[ancient hull]

Why does 11labs have voice verification now

[feral halo]

Ayo? @ancient hull level 2 !!!

[ancient hull]

Why do corporations hate fun so much

[chilly lake]

too many freeloaders?

[ancient hull]

I just wanna make a Skyrim mod man

[short crest]

Cock and ball torture

[haughty ibex]

where do i ask for help on making models?

[covert lake]

#📑│making-models or #✨│ai-help

[ancient hull]

Do you think it would sound as good if I used one of the premade 11labs voices and used an RVC model with it?

[honest saffron]

Anyone see "Computer Use" from Anthropic yet?

[covert lake]

i mean could work, dunno about the quality but u can test

[covert lake]

huh?

[honest saffron]

It's basically an A.I. agent that you watch work on your computer/browser.

Clicks through and inputs information. Nothing too crazy but I can see a lot of use cases and devs iterating upon it to make it do really neat things.

[feral halo]

Ayo? @honest saffron level 1 !!!

[covert lake]

ah didn't see it but seems cool

[tired jasper]

oh boy, after downloading all dataset, now i need to convert from ogg to flac 😢

[haughty ibex]

today is my birthday!

[tired jasper]

happy birthday!!

[amber light]

https://tenor.com/view/menhera-gif-6045874186498781362

Happy Bday

[elder willow]

https://www.weights.gg/chat/cm2jmjej60059nqz5fysrm80i?conversationId=cm2l0ow2l01d9nr6a8oul0s71

Happy Birthday!

[elder willow]

https://tenor.com/view/happy-birthday-gif-9122539521140949340

[fervent bone]

Is there a way to remove robotic voices in audacity?

[delicate stone]

wow happy birthday.....you want classic song for the day? dm me...

[astral crest]

Hi, everyone

[tired jasper]

finally training full overwatch characters beatrice

[elder willow]

@tired jasper could you teach me how to use ai

[tired jasper]

Each AI has own way to use, be specific

[brisk gust]

bro walked in and dropped the most open ended question on the planet

and then ghosts

[tired jasper]

training beatrice v2 with 41 voices 👀

[feral halo]

Ayo? @tired jasper level 10 !!!

[tired jasper]

beatrice is revolutionary, even for multivoice (supports more than 200 per model) it keeps the 23mb size

black magic

also runs realtime on cpu and it's very stable, while training is SO EASY, no transcription needs

i lost only .30its per second to train 41 voices at same time compared to one by one

FAST 🚀

[buoyant star]

@tired jasper I tried training with dataset in my native language, but the results were terrible. It's probably because it wasn't included in the pretrained languages. As far as I know, only English and Japanese were included.

[feral halo]

Ayo? @buoyant star level 2 !!!

[tired jasper]

Hmm, interesting, but you don't need to train in your native language, you just need to train in supported language and it will work for any other

But yet don't make sense for me since it doesn't use transcriptions to train

Part of my dataset is noise (non words) and it works pretty well

Also, another part of the dataset has a voice effect (not raw), and still works well.

Beatrice even auto adjust the tranpose

Maybe you tried in a very old version

[buoyant star]

I used a speech dataset that was 1 hour long

and it took 4 hours to reach 10,000 steps

[tired jasper]

I'm using a maximum of 600 (around) samples for each voice and each one with max 7 seconds and minimum around 3 seconds length

[polar flax]

for rvc or beatrice?

[buoyant star]

beatrice

[polar flax]

have you tried on singing, laughing, moaning, etc. samples?
few seconds for each voice seems like just zero-shot grade

[tired jasper]

Yes, even pig squeal and it keeps stable

RVC hallucinates a lot on pig squeal, beatrice just stay quiet

Beatrice is even better when using their vst3 plugin

RVC amplify the robot effect of autotune, i will test tomorrow what beatrice does when the input is using autotune

I bet it will be very good, i tried a Zenyatta sample as input once and it just "derobotized" it

[steel dome]

who tryna teach me all sub genres of trap and ungatekeep drum kits

[earnest dragon]

No Beatrice v2 doesn't have better quality then rvc it's just you think because you can run better settings easier on Beatrice but if you did for rvc and it worked then rvc would be better

[earnest dragon]

No rvc2 works good with autotune if you have a good trained rvc2 model

[open iron]

do yall recommend training on weights or RVC

[earnest dragon]

Yeah it's meant to make it not robotic rvc because the voice is what's robotic and replacing it makes it not have the auto tune sound but have the accuracy of autotune

[earnest dragon]

Depends can you change the Pretrained model

[open iron]

i guess not in weights

[amber light]

well the obvious answer is the official thing itself

[small cosmos]

Just got diagnosed with heroes

Herpes

[copper apex]

same

[feral halo]

Ayo? @copper apex level 2 !!!

[zenith hamlet]

Guys,is training a 7m dataset with 5000 epochs normal?

[covert lake]

5k epochs are way too much

its better u put 500 or 1k epochs and use the tensorboard, more epochs don’t mean more quality

[broken raft]

I don't think it's normal. Typically, 1000 epochs would be the maximum limit for training. I went over it before when training and the result sounded awful.

[tacit falcon]

i thought you just keep going until it rises no matter how many epochs it takes before loss stops going down

if the loss keeps going down its not done i thought

[earnest dragon]

Yeah that's how it should be you train until loss plateaus

[zenith hamlet]

That's what i said

[polar flax]

I'd suggest to check the tensorboard graph, it's usually around 200-500 epochs or may depend on dataset length

[elder willow]

.

[zenith hamlet]

I suggested that person that,they don't know what it is so i invited them here

[polar flax]

I have told him so there but he had no idea about tensorboard, until he could realize the graph may have gone up sharply at that much epoch

[vestal temple]

Komm pn

[kindred ore]

what quick question idk if this is stupid but does me talking and singing using rvc trains the model also

[feral halo]

Ayo? @kindred ore level 1 !!!

[covert lake]

nah doesn’t work like that

when u the model, it’s called inference, won’t train it tho, it’s 2 different processes

[kindred ore]

alright thank you

[grand tangle]

Hi! I'm...new?

[leaden meadow]

Is this GC active

[covert lake]

yw

[covert lake]

should be the lil man icon

top right

[limpid silo]

Hey can anyone help me whenever I try make my own voice model it sounds glitchy can anyone help me?

[covert lake]

whats ur pc gpu and what are u using

also, be sure to clean ur dataset

[limpid silo]

I’m using this https://colab.research.google.com/github/iahispano/applio/blob/master/assets/Applio_NoUI.ipynb

[covert lake]

yea that’s good

btw there is an ui version that could be easier

are u following a yt tut?

[limpid silo]

yeah

[covert lake]

send the yt tut

99% of yt tuts are old

[limpid silo]

Ok

[uncut tartan]

idk if someone can understand the voice changer but i don't btw

[limpid silo]

I’ll send it give me a sec

[covert lake]

I’m not a wokada helper, but what’s ur pc gpu and don’t follow yt tuts

[uncut tartan]

my gpu?

[limpid silo]

I’m using this

[covert lake]

yes, ur pc gpu

[limpid silo]

https://www.youtube.com/watch?v=MfjKeE_s7mE

[covert lake]

You can check your pc gpu via:
ctrl+shift+esc (task manager) -> Performance tab -> GPU

[limpid silo]

Im using this https://www.youtube.com/watch?v=MfjKeE_s7mE

[covert lake]

thats old and not even about the tool ur using

[uncut tartan]

the 0 or the 1

[covert lake]

what’s ur pc gou

[covert lake]

both

[limpid silo]

So what’s the updated one?

[covert lake]

I need to know ur pc gpu first

it matters alot

[limpid silo]

Ok how can I show u

[covert lake]

if you got a good pc gpu, u can do it locally (running on ur pc), else u have to use cloud (remote good pc) computing services like google colab and kaggle

[covert lake]

#🧬│ai-chat message

[uncut tartan]

RTX 4060 and an other one

[feral halo]

Ayo? @uncut tartan level 1 !!!

[uncut tartan]

intel R uhd

[covert lake]

yea the rtx 4060 is good enough

-rt

[rare sorrel]

This interaction has expired, use the command /guides if you wish to see it again.

[covert lake]

use the 1st guide

the wokada ( program for using rvc models for realtime calls) fork (modified version. This one got more performance)

and no there are no videos

[uncut tartan]

what should i download?

voice-changer-linux-amd64-cpu.tar.gz
402 MB
Aug 25
voice-changer-linux-amd64-cuda.tar.gz.aa
1.95 GB
Aug 25
voice-changer-linux-amd64-cuda.tar.gz.ab
1.05 GB
Aug 25
voice-changer-linux-amd64-rocm.tar.gz.aa
1.95 GB
Aug 25
voice-changer-linux-amd64-rocm.tar.gz.ab
543 MB
Aug 25
voice-changer-macos-amd64-cpu.tar.gz
308 MB
Aug 25
voice-changer-macos-arm64-cpu.tar.gz
190 MB
Aug 25
voice-changer-windows-amd64-cuda.zip.001
1.95 GB
Aug 25
voice-changer-windows-amd64-cuda.zip.002
556 MB
Aug 25
voice-changer-windows-amd64-dml.zip
457 MB
Aug 25
Source code
(zip)
Aug 25
Source code
(tar.gz)

[covert lake]

https://huggingface.co/Shadicti/deiteris-Fork/blob/main/voice-changer-windows-nvidia-b2309.zip

[limpid silo]

So how can I make my custom ai trained model not glitchy?

[covert lake]

its said in https://rentry.co/ForkVoiceChangerGuide#download-nvidia

please read the guide

[limpid silo]

?

[uncut tartan]

thx$

[covert lake]

That was a reply to a whole different thing. To help quality u need to clean better the dataset, but you need to tell me your gpu

please tell me your gpu name

ur using an old yt tut

[limpid silo]

My gpu is 11th Gen Intel(R) Core(TM) i7-11800H @ 2.30GHz

[covert lake]

That’s a CPU

[limpid silo]

Oh

Give me a sec

Ok here, NVIDIA GeForce RTX 3050 Laptop GPU

[covert lake]

could u tell me also the dedicated gpu memory

[limpid silo]

Ok

Wait..

GPU memory = 0.4/7.9 GB

[feral halo]

Ayo? @limpid silo level 2 !!!

[limpid silo]

?

[covert lake]

8gb

laptop gpus are weaker than desktop ones

u could try to do it locally (on ur pc)

[limpid silo]

I have it’s in my friends house tho

What’s the next step?

[covert lake]

As you got a good PC, you can use RVC locally, you can choose between:

• Applio: A fork of RVC with some extra features like Applio TTS, same quality tho
• Mainline: The original RVC

those are guides

[covert lake]

it isn’t ur pc ?

[limpid silo]

No bro that’s my laptop

[covert lake]

u can follow one of the guides above

[limpid silo]

Ok

[covert lake]

also check https://docs.ai-hub.wtf/rvc/resources/datasets/

[limpid silo]

Could u make a ai model for me?

[covert lake]

Sorry but nope, it takes alot of time

[limpid silo]

All good bro

[covert lake]

if u really don’t want to do it, #1159289738314919936 or #1191429836321849435

paid requests are done way faster than free ones

[limpid silo]

I have the money in my pocket how do I get a paid request one?

[covert lake]

either do a request with the paid commission tag in #1159289738314919936 ir contact a model master via #1191429836321849435

[stable sandal]

👋

[native saffron]

Where can I share some art at? 😎

🎶 this stuff is bombz

[covert lake]

#🏙│ai-images

[native saffron]

#1159290752195633273 is where I sent it idk lol .. is that chill?

[covert lake]

yea that, i thought u meant images with art lol

for music u should use #1159290752195633273 yea

[native saffron]

Ah lol.. word up

[feral halo]

Ayo? @native saffron level 1 !!!

Ayo? @novel violet level 2 !!!

[covert lake]

Oh yea

that to me too

ig

discord moment

[hidden ice]

hey

[native saffron]

||I flow magic||

[elder willow]

@atomic vector pls explain ggbbfgfddfg

[atomic vector]

it looks like you typed a bunch of random letters. can you please give me more context or tell me what you're trying to say? i'd be happy to help if you meant to type something specific
-# AI-generated responses may be inaccurate; please verify important information.

[native saffron]

That user name requires high level stats

[hidden ice]

^^

[native saffron]

I'm level 100 Eagle Vision

[rich sail]

Hello Guys I am Sahil Anand, a computer science student aspiring to be an ai engineer in near future

[hidden ice]

welcome 🙂

[feral halo]

Ayo? @hidden ice level 4 !!!

[hidden ice]

gg

[rich sail]

who are also preparing for the same profession or in this field already?

[native saffron]

I'm a firm believer

[polar lava]

hey guys when I want to update my GPT, I get this error:

Model o1-preview does not support custom system messages

any idea how to solve this?

[hidden ice]

hey

[ionic pumice]

https://cdn.discordapp.com/emojis/837763484787671050.webp?size=48&quality=lossless&name=rioluok

[tired jasper]

Idk why you insist that RVC is better, in my tests beatrice was much better

It's not just train in real data?

[glass jackal]

the problem im running into are clients dont want to train on their live customers esp. in ecomm because its mission critical. if your voice agent disrupts their revenue or conversion rate, that would be a problem. so they currently have us and their internal team stress test it and take forever to deploy into production

[serene jacinth]

Hello

[narrow mountain]

the new sd 3.5 8b 15gb came out is it good or meh ?

[crimson steeple]

Hey anyone know where I can find an ai voice of shadow, sonic, and mephiles?

[solar torrent]

#🔍│find-models or #1175430844685484042 right there.

[feral halo]

Ayo? @solar torrent level 12 !!!

[crimson steeple]

Thank tou

[tired jasper]

so i have no idea how to help you

[crimson steeple]

Hey anyone know what I do to run the ai voices?

[lament mulch]

does anyone here have okada working with a very low delay / sample size? dm me

[feral halo]

Ayo? @lament mulch level 1 !!!

[covert lake]

what's ur pc gpu?

Also are u looking for:

• training (making rvc models)
• inferencing (using models) on pre-recorded audios
• inferencing on realtime for calls
  ?

[crimson steeple]

I want to edit the ai voice over actual dialogue

[covert lake]

#1159290752195633273

[crimson steeple]

Like ai voice over a song for example

[covert lake]

so, an ai cover?

[crimson steeple]

Yeah

[covert lake]

what's ur pc gpu?

[covert lake]

You can check your pc gpu via:
ctrl+shift+esc (task manager) -> Performance tab -> GPU

its the one at the top right

[crimson steeple]

Sure thing I'll check it out later today

[covert lake]

alr, are u on mac perchance?

[tired jasper]

apparently multivoice beatrice is a mistake

don't worth, my solo voice is superior

[loud dirge]

les différentes définition sur le management foncier avec références bibliographiques

[tired jasper]

english please

[final vortex]

Hello.

[crimson steeple]

Windows

[covert lake]

alr good, then follow what i said above

[covert lake]

#1159520446799155350

[solar torrent]

English only in chat please.

[small cosmos]

My dick so fucking hard

[feral halo]

Ayo? @small cosmos level 1 !!!

[solar torrent]

Just make it soft.

[neon ore]

Help how do install and use the client am lost af

WITH A YT GUIDE

[solar torrent]

Tutorial videos on YouTube are kinda outdated for the Okoda and RVC. There's a tutorial document for them available somewhere.

[sweet garden]

hi

[neon ore]

[wind stratus]

-guide

[rare sorrel]

📚 Documentation

AI HUB Docs

https://docs.ai-hub.wtf

🍏 Applio Docs

https://docs.applio.org/

✨ More guides

• How to use RVC Mainline Colab by Cauthess
• AICoverGen Colab Guide by Eddy (Spanish Helper)
• Create a model with RVC disconnected (colab) by Angetyde
• Italian Guides by Ilaria

[solar torrent]

-okoda

[neon ore]

lulu has lost it

[feral halo]

Ayo? @neon ore level 1 !!!

[covert lake]

all youtube tuts are outdated

do NOT follow them lol

what's ur pc gpu?

[neon ore]

i7-10400kf

[covert lake]

that's a cpu

You can check your pc gpu via:
ctrl+shift+esc (task manager) -> Performance tab -> GPU

[wispy dock]

How does this AI art process works?

[neon ore]

oh mb

misread it

NVIDIA GeForce 4060

[solar torrent]

Each AI generator like Midjourney, Stable Diffusion and Dall-E has its own method. But for the Stable Diffusion one, it generates from noise.

[covert lake]

should be good enoug to locally

-rt

[rare sorrel]

This interaction has expired, use the command /guides if you wish to see it again.

[covert lake]

1st guide, wokada fork

you needa read it up

[wispy dock]

Twitter

[dusty glacier]

You mean 𝕏??

[stray linden]

hello im jordi

can someone help me with chris cornell ai voice in visual studio

how can i use it?

i have the zip from this page https://huggingface.co/QuickWick/Music-AI-Voices

https://huggingface.co/marcoc2/so-vits-svc-4.0-models (this)

[covert lake]

you need to use the program rvc

[covert lake]

that's so vits svc, its very outdated and lower quality

[covert lake]

most of those voices are rvc

what's ur pc gpu?

[stray linden]

and there's one with better quality?

[covert lake]

rvc is the best program since a year

[stray linden]

rtx 3050

[covert lake]

how much vram?

like gpu memory

[stray linden]

4 GB GDDR6

[feral halo]

Ayo? @stray linden level 1 !!!

[covert lake]

4gb gpu memory (aka vram) are very low

i don't think u could be able to even inference (use models) locally (on ur pc)

[stray linden]

why? It needs so much?

[covert lake]

You should use cloud (remote good pc) cloud computing services

[covert lake]

AI takes ALOT of computing, plus 4gb vram is shitty nowdays

for basically everything

[stray linden]

do u mean google cloud?

[spring warren]

Are there any google Collab links that work

[feral halo]

Ayo? @spring warren level 2 !!!

[covert lake]

ud need atleast 8 gb of vram to make it work

[covert lake]

well yea, there's google colab, kaggle, zerogpu huggingface spaces, lightning.ai

the fastest and free one is zerogpu huggingface spaces such as ilaria rvc zero, zerogpu offers A100 gpus

[stray merlin]

Ilaria RVC: CLICK HERE 🤗

Guide on how to use it: CLICK HERE 📝

Don't forget to thank Ilaria if you find it useful! 💖

[stray linden]

but how can i use with the chris cornell voice?

[covert lake]

for what?

[stray linden]

you know how can i used?

[covert lake]

yes, read the guide above

[spring warren]

Yo boss any idea for google Colab link that works , or an alternative

[stray linden]

there's no guide i think

[spring warren]

To make ai covers that type of stuff

[covert lake]

blue text = hyperlink = text that if u click it it redirects u to the site

[stray linden]

okay thanks, im gonna prove it

[stray linden]

and sorry ahha where is this?

[covert lake]

alright, that's also called inference (use models) on pre-recorded audios, if you really want to use google colab, u can use Applio Colab, but i'd suggest u to use the ZeroGPU (A100, much more faster than the T4 gpu that google colab offers) HuggingFace Space Ilaria RVC Zero

[covert lake]

the message above #🧬│ai-chat message

[spring warren]

Thank you boss , are the steps still the same to make those type of cover audios

[covert lake]

You're welcome, should be pretty similar and easy, you have to read the guide tho

[stray linden]

thanks a lot

[covert lake]

your welcome

[wispy dock]

Interesting

[chilly lake]

all diffusion type models start with random noise

even RVC starts training with a random noise

[worthy coyote]

why do i hear hl2 sound effects?

[neon ore]

-rt

[rare sorrel]

This interaction has expired, use the command /guides if you wish to see it again.

[covert lake]

#1159290752195633273

@swift beacon

[swift beacon]

oh

[tired jasper]

in the room with GPU on training: 🔥 🥵

trying again to train beatrice in multivoice but this time i set 30k steps

tried to continue a 10k model, but got some unstable behavior with loss function

seems like continue a finished training is bad idea

or maybe the model just can't go beyond 10k steps

i will discover it soon

for normal human voice i feel it ok to multivoice

but for characters that have voice effect it's not good at 10k steps

[fathom garnet]

[terse otter]

[spiral timber]

You guys believe in AGI very soon?

[elder willow]

hello

[polar flax]

no I believe in INT

[spiral timber]

What is that

[toxic yoke]

anyone know hoy i can use applio in python?

im creating a assitant and i need a tts

[raven bough]

hello

[feral halo]

Ayo? @raven bough level 1 !!!

[raven bough]

Does anyone know where I can find how to remove audio silence in Audacity?

[wind stratus]

truncate silence

[raven bough]

what is the best configuration?

[wind stratus]

i got this config from previous sessions, it may be a lil harsh but its good

[south cave]

is anyone here good with UVR 5 stuff?

[valid gyro]

1

my name is Jaroslav

hello

[south cave]

hello

[open iron]

FUCK mainline I'm never using it again, it fucked up the PC screen and is black now

Dammit

[earnest dragon]

It wasn't mainline is was you messing up

[open iron]

Probably

[earnest dragon]

No not probably it was a definite you

[open iron]

Yeah..

[earnest dragon]

What gpu

[open iron]

1060

[marsh nexus]

yo

[open iron]

nvm my pc just gave me a scared and then the screen came up

what a scare i had

[covert lake]

#1159290752195633273

[south cave]

blue screen of death?

or the buffering screen of death?

[open iron]

no the black screen

[feral halo]

Ayo? @open iron level 12 !!!

[south cave]

oh god

[open iron]

no mouse cursor anything

it was pure black

[south cave]

thats a more worse fate than any horrible pc fate I've heard

[open iron]

it fixed himself

[south cave]

oh

[brazen marsh]

t

[chilly lake]

driver crash as it ran out of memory?

[buoyant radish]

Hey folks, is Applio down for anybody else?

[leaden raft]

bro i cant use applio without erros

[feral halo]

Ayo? @leaden raft level 2 !!!

[leaden raft]

for me

its just errors

bro nothing about applio works

the report a bug thing doesnt work]

literly nothing its all broken

[mossy heron]

jo guys in uvr are wich mdx models do i need again and in wich order (didint use ai fore quiet some time )

[signal garden]

Hello guys, Juls here. starting as a newbee trying to generate some RVC. Is there an easy way to filter voice-models by language, like for instance using a special tag in the search bar ?

[amber light]

in the list, yes there are tags

there are 8 language options/tags you can filter from

[amber light]

#✨│ai-help

[leaden raft]

ty

[broken vapor]

oy

[feral halo]

Ayo? @broken vapor level 1 !!!

[tired jasper]

realized that Beatrice requires some skill from user

it's not just use

you need to interpret character

i tried to record the video explaining twice, but got no video

OBS refusing to record my screen

i really need to start testing before records

[harsh drift]

hola