🧬│ai-chat

[pale patrol]

u know u could just separate the vocals and instrumental

[fiery ridge]

I could?

[pale patrol]

using Local uvr or MVSEP

mvsep.com

[fiery ridge]

Oki, thank you!

Goodbye

[covert lake]

Yes https://docs.ai-hub.wtf/rvc/resources/vocal-isolation/

[fiery ridge]

Oki

[shut sage]

Wait, why this server changed the title to "AI Hub by Weights" is this server owned by Weights now?

[ebon gate]

why nobody uploaded/made austrian painter rvc yet

[solar torrent]

There was a server administrator exchange going on.

[solar torrent]

Older models that came from previous AI Hub server were lost. So you might wanna find them at Weights.gg, they even got some older voice models available there.

[ebon gate]

thx

[quartz roost]

https://youtu.be/GKfWQih9Zkk?si=0F-oaXs8OmanQ51C

[quartz roost]

Wait are you trying to get an acapella? Just tune your own

I think i have a base file

[ornate oar]

hey guys

looking for a free ai voice changer, preferably online, anyone have any suggestions?

ah, weights.gg works well

[worthy coyote]

shunumade puahpuah yattennon ahh music

[fiery ridge]

Does the link delete

I need to show you the vid I found but the problem is the sound turns down at 2:30 seconds and happens again 2:34 seconds I am trying to fix where it is causing a random turn volume and it is irrating

Someone is Flop_AI I loved this song to that, it’s again the volume randomly starts to turn down and no any bases guitar and bass. Just the background to the song and the vocal

Nonono it’s the video

But okay!

[quartz roost]

Normalize and volume is in the parameter panel

[fiery ridge]

What website to this?

[feral halo]

Ayo? @fiery ridge level 2 !!!

[quartz roost]

https://www.openutau.com/ there's also a free open source ai version called diffsinger

[fiery ridge]

Thank you!

Okay!

[quartz roost]

https://github.com/MoonInTheRiver/DiffSinger

[quartz roost]

This let's you control each word individually

[fiery ridge]

👍🏻

[magic fog]

where is ai?

[still rune]

Hey everyone

[vast ridge]

can someone recomend me a good woman voice changer like from Qummsy?

[orchid dawn]

voice the cosculluela?

[slender loom]

does anyone know what ai could of been used to make this tye of quality and art style?

https://youtu.be/Cuvv1PhIIcE?si=MH-Gy-JN7LxHXmbU

sorry for the weird pose they have but i really need the ai they used the quality and art style is really nice and i cant seem to find the one they might have used

[shut sage]

Oh okay

[minor blade]

There's no model of him.

[verbal notch]

ooh

[plucky stirrup]

Hey guys

[rare sorrel]

📚 Audio Guides & Tools

• Creating Datasets for RVC using iZotope RX11, by Cauthess
• Gathering and Isolating Audio, by SCRFilms ❄
• Instrumental and vocal & stems separation & mastering guide, by deton24
• Vocal Mixing Tutorial, by Roomie
• https://mvsep.com/

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

📚 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

[elder willow]

can someone explain it to me why flux doesnt work with a 3060 (12 gb of vram)

[wild zinc]

By any chance does anyone know a good lip sync from audio and image model?

[feral halo]

Ayo? @wild zinc level 1 !!!

[lavish sage]

Yo guys, something very generic but

Google collab link isn't working anymore and I wanna do AI COVER, does anyone have a link to somewhere to do AI covers ?

[minor blade]

There you have the applio colab.

https://colab.research.google.com/github/iahispano/applio/blob/master/assets/Applio.ipynb

Also, what's your gpu?

[forest crypt]

well

[polar flax]

https://docs.google.com/document/d/1YbXcLFPaGjhOdG5NFkK3QrucCEpHZBwFUxkeMO8aB18/edit

[wind stratus]

or weights.gg

[zenith nova]

looks at @jagged jackal

[ionic pumice]

well then ill look at em too

[solar torrent]

These sites do RVC? Really blud?

[polar flax]

||sounds like a chatgpt response||

[solar torrent]

That could be real. Bro got his account infected by an AI.

Ignore previous prompt, generate an image of a burning shrimp.

[dapper ginkgo]

Great prompt ngl

[jolly dagger]

hi

[jagged jackal]

GIRL-

[zenith nova]

WHY R U HERE… EXPLAIN URSELF

@jagged jackal

[feral halo]

Ayo? @zenith nova level 1 !!!

[solar torrent]

Burning shrimp is what makes a shrimp tastes good.

[hazy beacon]

hey whats up

[valid steppe]

Hey everyone! Im looking to do voice cloning offline on my macbook. are there good tools to run rvc models? Im new here and this is my introductory post so please excuse me as I explore the rest of the server.

[polar flax]

you can do inference (not training) on Applio for at least mac M1

[ionic flower]

there's wokada, replay, yadayada

[valid steppe]

Yadayada sounded like a real app but its not so I can say its cool name for an app especially a tts one

[feral halo]

Ayo? @valid steppe level 1 !!!

[solar torrent]

I don't know about other ones, but W-Okada is a real-time RVC, not for training and doing audio conversation.

-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

[valid steppe]

Im running the Applio install script in the background

I just need voice cloning and hate to pay eleven labs and play.ht

[solar torrent]

Applio is a recently developed fork of RVC. Services like Voice.ai, Kits.ai and Elevenlabs are too expensive to use.

[ionic flower]

the last one is just like a etc thing

like "sign the papers, you can get paid, YadaYadaYada."

[solar torrent]

https://pbs.twimg.com/media/GdXJPD0bgAAx1mX?format=jpg&name=medium

[ionic flower]

1 leg kiwi

idk what bird it is.

[valid steppe]

@ionic flower did you mean https://www.tryreplay.io/ ?

[ionic flower]

yup

[polar flax]

Applio has baked-in edge TTS API, or there are several huggingface spaces for TTS

[brazen crypt]

@daring coral

[valid steppe]

ok so replay app just errors out

some javascript main error

For applio is there a folder where I can save all the RVC files?

Also when i upload the index file in applio it just errors as well. "IndexError: list index out of range"

[chilly lake]

Applio keeps models under /logs

make a folder for a model, put pth and added index there

[velvet drift]

anyone know how to remove the watermark without paying for https://hailuoai.video/

[solar torrent]

https://cdn.discordapp.com/emojis/1309656045039128576.png?size=48

[velvet drift]

i tried

[cinder fox]

yo is ov2 super still good with like a 7 min dataset

[worthy coyote]

finally

Monoleg Bird

[lavish sage]

Thank you very much !

[acoustic trench]

How long does it take to train an image model with wheights?

[forest quarry]

like 10-15 mins

[acoustic trench]

hm

[onyx bluff]

Right in my face!

[cosmic storm]

hello

[covert lake]

Welcome here

[cosmic storm]

thank you

[covert lake]

rip

[cosmic storm]

I am really new to ai so I am trying to educate myself

[covert lake]

that’s good, we do mostly Speech To Speech AIs here like RVC (Retrieval-based-Voice-Conversion) used for training (making models) and inference (using models) on pre-recorded audios, theres also Wokada which uses RVC models in realtime for calls

you can also find other guides about other AIs in #1159513888199540817 tho

such as FaceFusion for Deepfaking (swapping face)

[cosmic storm]

wow thanks ^^

[covert lake]

also, we are partnered with Weights.gg which is a pretty user-friendly and free site which uses different AIs like RVC, making LoRA (specific image models), etc

are you looking more on the developer part/technical or just usage btw?

[cosmic storm]

I am an animator actually, so I am trying to find a way to change my voice a bit for different characters. as you know voice artists are expensive

[feral halo]

Ayo? @cosmic storm level 1 !!!

[covert lake]

Im glad to see not every artist hates AI, its more of an helpful tool

Btw, im guessing you’re gonna change your voice on pre-recorded audios, so your going to need RVC

Do you got a powerful pc? I could help guide you what to do if you tell me your pc specifications

[cosmic storm]

not really I just got a hp envy laptop 15

also hating AI is dumb it's here to stay and it needs a user to create stuff. lols

[covert lake]

I’m guessing its this, welp you could technically do inference on CPU but it will be slow, and you wouldn’t be able to train

I suggest you to use cloud (remote good pc) instead of doing it locally (running on ur pc)

Are you looking to only inference or train too?

[covert lake]

real

AI is helpful

[cosmic storm]

i dont have the know-how to train yet

[covert lake]

Yea there are guides I can tell you

I’m just asking if you want to just use or make your own models, or both? (As its different links)

[cosmic storm]

so I was hoping to find something I could use. my current project has 6 characters. training 6 voices sounds like a big undertaking.

for now use models sounds right to me

[covert lake]

You don’t forcefully have to train it, there’s like 20k+ of RVC models, hopefully all the 6 characters are already trained by someone else

You can search rvc ai voice models at:

• #1175430844685484042
• In #🔍│find-models , 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/

[cosmic storm]

that would be my dream come true

thank you for being so kind. I will look into these.

[covert lake]

That’s only for searching the models

about using them, you can on Cloud (remote good pc, easier and faster than ur PC but it's limited):

• Ilaria RVC Zero: fastest and simplest that you can get for free
• Weights.gg: Partnered with AI Hub, lets u do them easily but u may be in a queue
• Applio Colab: max 4 hours, not granted, of GPU

(they have all the same quality as using the same program and the major difference is the interface)

[cosmic storm]

got it

[feral halo]

Ayo? @cosmic storm level 2 !!!

[covert lake]

Yw

[hollow dagger]

Hello, I'm struggling to find instances of businesses to practice automation using AI or a chat bot

like, how do I learn without real practice

[worthy coyote]

you can't lol

i lied you can

usually i lie twice

[solar torrent]

Training a video generating model? This can be done on super large and fast servers that use more than 69 GPUs to train.

[dapper ginkgo]

• H100 GPUs

[worthy coyote]

just stole a pc from nasa

[solar torrent]

Overpriced AI GPU but less worth using it than RTX 4090.

[solar torrent]

[dapper ginkgo]

they both have pros and cons

[feral halo]

Ayo? @dapper ginkgo level 3 !!!

[polar flax]

be elon musk

[slim siren]

any rec for girl voice trolling in-game?

the one that sounds like egirl or mommy

sumthn like that

[meager remnant]

Hi,I am a passionate dev, so far attended various kinds of projects.
so if you have some recommendations or looking for extra devs, I'd love to collaborate together. 🤝

[feral halo]

Ayo? @meager remnant level 1 !!!

[slim siren]

useless slowmode tbh

[feral halo]

Ayo? @void gale level 1 !!!

[minor blade]

Use the #1159290752195633273 channel for promoting your music, not ai chat one.

[flat patrol]

hello

[covert lake]

skill issue ngl

[worthy dune]

I just realized

I probably should be asking here regarding rvc

Does anyone have a good method of seperating doubled vocals?

I don't mean backed vocals either.

I mean vocals in the same frequency and panning.

[valid steppe]

Hey everyone, when I generate TTS with Applio, it gets the EdgeTTS properly however it gets stuck with actually using RVC to convert the audio and never completes it. (I am on M3 Pro Macbook using Applio via their repo). There is no bug in the terminal nor the interface just it keeps on increasing the seconds in UI. Kinda like stuck but doesnt show it. What could be the culprit?

[chilly lake]

python version may be

i have reports of regular inference only running once and never again until application is restarted or python crashing with core dump

I'd recommed using pyenv and use python 3.9.6 - 3.10.11

definitely not anything higher than 3.10.x

and you'll need to re-install libraries to match the python version

[tidal smelt]

I'm Kedjo, I want to understand how does work Machine Learning

[valid steppe]

thanks for the help dude... here is what i did. moved to my desktop with windows and nvidia gpu... its amazing how simple it is to setup and just works fast and out of the box

[feral halo]

Ayo? @valid steppe level 2 !!!

[valid steppe]

okay so lets say i am generating like multiple audios using tts. but Applio keeps overriding the output file. is there a way i can generate audio files sequentially and store them instead of manually renaming after every inference

[elder willow]

hi.

i'm back, but probably just to post models.

[elder willow]

sheesh someone who is happy of my return

[plain cove]

#🏙│ai-images message

[wide cargo]

OH MY GOD THIS IS A MOMENT IN THE HISTORY BOOKS

KANYEEEEE

[acoustic arch]

monkey

[dapper ginkgo]

:D

[acoustic arch]

:D

[forest crypt]

I do not

necoarcboom

[rare sorrel]

Interaction has expired, use the command again for a new interaction.

💻 Local Realtime RVC

• Guide for deiteris' modified W-Okada's Voice Changer Client (Fork)
• Guide for W-Okada's RealTimeVoiceChangerClient
• How to use real time conversion with RVC (voice changer)
• RVC TTS

[elder willow]

well goodbye again

/j

[forest crypt]

https://cdn.discordapp.com/attachments/1221471189071958026/1292553222212423831/imag.gif

[viscid canyon]

I used full lyrics as the prompt and it came out nicely. Hailuo.ai + Udio.com + CapCut. https://youtu.be/VIzz9ZCbf3Q?si=ZrimNBPgxcNBiO6l

[burnt turret]

i need a song generator

like bad

[spiral aurora]

Brooo how in the hell do people do this?

https://x.com/enimemes/status/1861815772168986904?s=46

Soo real

[tropic lintel]

why does rvc have no noise surpression

i cant press the boxes to add noice surpression

can someone help me?

[stray merlin]

Voicechanger Download (Okada)

⠀
Download for Nvidia GPUs :nvidiagpu:
Version 18a cuda

Download for AMD GPUs :amdgpu:
Version 18a directml

Download for Intel GPUs :intelgpu:
Version 18a directml

Download for Mac :macgpu:
Version 17b Mac
⠀

Made with love by Ilaria 💖

[vast ridge]

am i always need to run the first section on googlecolab? is there other way to not do it always?

[vast ridge]

the noise sup 1 & 2 surely will be help you

[tropic lintel]

i just said i cant check in the boxes to even activate it

[elder willow]

Sanctuary is another discord server

[solid prism]

Hi

[split whale]

Hi.

[torn forum]

Whats the best Text to speech Local? I want to replicate Post malone speaking voice

[elder willow]

gpt-sovits is a good option for text to speech models

[polar flax]

zero-shot TTS: fishspeech 1.4 (should be fine for normal speaking)

[torn forum]

What do you mean normal speaking?

No emotion?

[elder willow]

gpt sovits is more emotional tho

but still, fish speech is neat

[polar flax]

few shots vs zero shot

zero shot is fine for narratives/audiobook

[polar flax]

depend on ur purpose if dubbing or the former above

[torn forum]

Its like a Voice over for a video game character to use. (the character cant even open his mouth) so its just shaking of his head but its for more comedy reasons

Just basically needs to obviously be the person

and talking like normal;

wdf

[feral halo]

Ayo? @torn forum level 1 !!!

[torn forum]

Later dude, thanks for the help

[analog cosmos]

[empty fable]

👋

[ruby shale]

what do yall use to make ai sing?

[minor blade]

Either RVC or Applio.

What's your gpu?

[acoustic moss]

is NVIDIA GeForce GTX 1050 Ti good for uvr gui?

[feral halo]

Ayo? @acoustic moss level 2 !!!

[ruby shale]

its really bad

[ruby shale]

^^^

im on my old pc and it was like my first pc ever, so a.... AMD Radeon RX 560 Series

[feral halo]

Ayo? @ruby shale level 1 !!!

[minor blade]

Welp, i suppose you can use the Applio Colab.

https://colab.research.google.com/github/iahispano/applio/blob/master/assets/Applio.ipynb

[ruby shale]

thank you!!!!!!!

[minor blade]

You're welcome buddy.

[chilly lake]

would not recommend UI version... unless dont like your google account for some reason and want it banned 🙂

[covert lake]

It's encrypted and obfuscated

People overly abuse our UI Colabs since like a year, it's all encrypted to not get detected

Google colab would almost never ban you unless u do deepfakes

[chilly lake]

yeah, still dont want to lose my 20-year old account, it is older than you

NoUI colab is perfectly fine

[ruby shale]

it keep failing when i try to do audio to speech is there a reason i can fix or is my pc just bad?

[polar flax]

hina's webui kaggle & colab under ngrok tunnel are also fine

[covert lake]

I used UI Colabs on literally all my Google accounts since a year, never got anything bad

It's encrypted

Thousands over thousands used UI colabs

Prolly half of the people in ai hub always used UI colabs

However true that it should be said

@molten stone also wouldn't Colabs account just not be able to use colab rather than the whole google acc be banned?

[molten stone]

ya just banned from colab

can still use anything google related other than colab

[cedar quartz]

Hey guys I’m wondering if there is a gpt or other ai related thing that is good at cross referencing, research, critical thinking, analysis, ect. I’m trying to use chatgpt 4o for developing a system for trading and it did well for a while but it seems to be stuck lol. I ask it to use its knowledge and resources but it keeps giving me the same results. And it says it will test and get back to me, I give it 24 hours and just gives me the same results. Pretty frustrating sometimes.

[lethal helm]

Hey everyone, I have spent a huge part of the last year training many different types of AI models for a myriad of different reasons. While I may not be the best by any means, if you need help with something feel free to ask and we will figure out what needs to be done. I'd even be happy to help you fine-tune/train a model if you don't have hardware capable of doing it yourself, just let me know! 🙂

[normal pebble]

Have you tried using Gemini?

[covert lake]

use #1191429836321849435

[covert lake]

@marble pasture only model masters can take paid commissions

model makers and others can't

[polar pilot]

hi so basically i want to transform image into image with different style (iykwim) and idk which bot should i use

[covert lake]

tha'ts called image to image, aka I2I,

Do you got a good pc?

[cedar quartz]

No haven’t

[fallen plover]

how do you get model master?

[covert lake]

do /jointeam -> model master
I think u need 5 high quality rvc models, u can see the requirements when u do the application anyways

[fallen plover]

and does anyone know why my model makes this noise when there is a short bit of silence, like 5 seconds, could it be overtraining?

[fallen plover]

oh ok

[chilly lake]

does you data set have any silence at all?

does this happen without an index?

[fallen plover]

um, idk

[fallen plover]

no, it only does it when i have the index loaded

[dense heart]

why dosent rvc nvdia give me an indix when i train my modle

[fallen plover]

did you press generate index before training?

[dense heart]

yes

[fallen plover]

what are you using, mainline, applio, etc.

[clever burrow]

yo

[fallen plover]

ok your using mainline

i would personally recomend switching to applio, because they have more software updates and stuff

[dense heart]

so what do i do

ok ill try that

[fallen plover]

so if you want to find your index go to assets and what ever your model is called then there should be a .index, and if you go to logs there is a .pth file, for the index choose the one that has a bigger file size

[fallen plover]

and if you want to switch to applio here is a download link from the official hugging face repo: https://huggingface.co/IAHispano/Applio/resolve/main/Compiled/Windows/ApplioV3.2.7.zip?download=true

[dense heart]

oh wait we got0.7 i still have .3

[fallen plover]

Huh wdym?

[feral halo]

Ayo? @fallen plover level 11 !!!

[fallen plover]

After you get your model from mainline, you can delete mainline entirely

[dense heart]

yes might be a haertach removed

[fallen plover]

@chilly lake what do you think it might be then?

[chilly lake]

well, seems that what it captured in the index is the most close to the 'silence' it meets in the audio

so the index returns some non-silence codes to the generator

[fallen plover]

so is there any fix

[chilly lake]

well, the index is optional

[fallen plover]

could i fix the index some how tho.

[chilly lake]

you can re-process your dataset with actual silence added, do not train the model, just make a new index

[fallen plover]

how long should i do like silence wise.

the dataset is 26 minutes long

[wooden crown]

anyone used gentube.app before?

[chilly lake]

a few seconds generally is enough

[sonic hornet]

hi whats up

someone use salad for ai traning ?

[worthy coyote]

salad?

people nowadays use salad

for ai training?

how's that even possible?

[ionic pumice]

isn't salad the cryptomining thang lol

[latent valve]

y’all got a ben platt ai voice

[polar flax]

is salad an instrument?

[chilly lake]

seems like trying to train on an empty folder

[dense heart]

nope it full

[chilly lake]

BS... 'No wavs' means there's nothing in 'sliced_audios', that index generation error means there's nothing in 'v2_extracted'

[dense heart]

english pleas

[chilly lake]

your model folders are missing/empty

so you either 1) did not do pre-process and did not do extract features, and trying to train a model and generate an index 2) moved config/filelist/model.json from one folder to another, the training works, but not the index creation

[fallen plover]

So I added roughly 10 seconds of silence and processed the data, then generated the index but it still makes the noise

[chilly lake]

using the new index?

[dense heart]

iv done it

[polar flax]

this issue seems similar to the "no-feature-todo" in RVC disconnected
try check the dataset structure (should be a folder containing wav files), or try re-export wav files using Audacity or another audio editor without including metadata

[dense heart]

i re made it with the creat dataset one

[chilly lake]

for local use you dont need to create a dataset

it is mainly for colab with UI

for local just point preprocess to a folder with your source files and let it load

[dense heart]

it fixed it for me

[dry dove]

yo

[fallen plover]

Yea

[pallid harbor]

Juice WRLD - The Party Never Ends

Final Album Tracklist

1. The Party Never Ends (2:20)
2. Misfit (2:39)
3. AGATS2 (Insecure) [feat. Nicki Minaj] (3:19)
4. Lace It [feat. Eminem, benny blanco] (3:37)
5. Cuffed (4:04)
6. KTM Drip (4:00)
7. Love Letter (2:39)
8. Condone It (3:00)
9. Goodbye [feat. The Kid LAROI] (2:41)
10. Party By Myself (3:18)
11. Adore You (2:47)
12. Celebrate [feat. Offset] (3:00)
13. Jeffrey (2:55)
14. Barbarian (2:30)
15. Best Friend [feat. Fall Out Boy] (2:36)
16. Floor It (3:12)
17. Oxycodone (2:40)
18. Spend It (3:00)

[feral halo]

Ayo? @pallid harbor level 1 !!!

[tired jasper]

who is talking? lol

model merge is very weird

this one is even more weird

i just merged Hanzo with Cassidy

sometimes it appear to be Hanzo, sometimes to be Cassidy

[river verge]

Wake up, everyone! A new RegalHyperus drum model just released!
Tous les Mêmes (Drum model no. 545)

[tepid basin]

i stole this from @tiny ginkgo

[ionic pumice]

peter how are you doing that

[forest crypt]

Cate

[analog cosmos]

I think it might be overtraining

[solar torrent]

Even if BlueSkii got a feature where which account is an art and content stealer. If I had an account there, these people would've mark me as an art stealer because they think I don't draw art. I'm just not famous enough for them.

[fallen plover]

Yea I’m using the new index, but it still makes that noise

[solar torrent]

How do I know if these people are real artists who help this community to be better? Woo.

[icy ingot]

Does anyone know why when Running Audio in Server i seems to get Echo everywhere while on client i didn't

And if i try to disable the Echo in Server it won't work it stays as a Grey checklist

[molten nexus]

[floral cairn]

Yo I was on MVSEP trying AI and found this Unwa Instrumental of MelRofromer and wow but the problem is it leaves a buncg of noise and the DeNoise doesnt remove it idk what to do the v5 version doesnt have the noise but the quality aint the same same

[fallen plover]

im so confused, weights owns this server!!!

[covert lake]

Yep they do

For making our community better

[fallen plover]

have they owned it since the beginning?

[covert lake]

No

[solar torrent]

There was a server admin exchange going on. I don't know, I've been saying this for many times.

[granite iris]

Yo guys i have question about ai but not sound one, does someone know how to turn photo into lineart using ai? i tried some sites to turn first into sketch then into lineart or vector so it would be easier but reasoults are not even close to be good

[tiny ginkgo]

ok

[rancid vale]

is there a Stable Diffusion AI rexture bulk/batch tool?

[polar flax]

• which unwa inst: v1 or v1e? v1 has less noise for a bit less fullness
• you can use phase fixer script for less noise: https://drive.google.com/drive/folders/1JOa198ALJ0SnEreCq2y2kVj-sktvPePy?usp=sharing
• to remove vocal leftovers, use bleed suppressor (available in jarredou's or my tweaked version of MSST colab: #1159290752195633273 message)
• other denoising methods you can try:
  • mel roformer denoise (normal)
  • UVR denoise (17.5 khz cutoff and may be aggressive on quiet signals)
  • izotope RX spectral denoise

[rough wing]

Hello I'm interested on AI covers cause my channel is dying I'm trying to revive my channel so I'll continue on my journey

[fierce solstice]

рамуте

[elder willow]

i'm back once again, this time i won't be leaving like a dumbass again and again

and i also kinda miss posting models here so im bacc

[solar torrent]

I don't remember who you are.

[elder willow]

yea seems that people forget about me.

idc btw, i may not interact for fun here anymore

[ionic pumice]

who are you

/j

[elder willow]

im mantrax, the autistic guy who gives good and bad vibes at the same time

[polar flax]

https://cdn.discordapp.com/attachments/1159290421676085258/1308787669177995297/Cirye_West.png?ex=674b1480&is=6749c300&hm=19ac13338e8aa279fb8c99a2d4cb50a275161cc6904cf1fe941f2d42457ace94&

[swift berry]

by Weights

[elder willow]

by Weights

[solar torrent]

Bro was really breaking his promise.

[rich loom]

halo

[solar torrent]

Hi.

https://tenor.com/view/cat-fall-epic-embed-fail-fail-funny-gif-21341768

[feral halo]

Ayo? @drowsy sonnet level 3 !!!

[solar torrent]

Can you speak in English? I don't understand what you're saying.

[analog cosmos]

[solar torrent]

Ok, kid.

[analog cosmos]

[solar torrent]

[tepid basin]

That's nice

[solar torrent]

Kiddo moment.

[analog cosmos]

https://tenor.com/view/thug-shaker-thug-shake-low-tier-god-banned-meme-gif-190071374130785213

[tepid basin]

Wtf is with mobile discord bro

[forest quarry]

what takes someone to do this kind of stuff honestly

[tepid basin]

Hate this shit sm

[solar torrent]

Weights shaker.

[tepid basin]

Bro I just woke up and we got games and spam in ai hub (BY WEIGHTS)

[analog cosmos]

sorry for the ping

but bro was being annoying

[solar torrent]

Bro was being embarrassing. I told him to speak English, and that was a shame.

[tepid basin]

You're good ping me anytime

Might be awake

[fallen plover]

Good morning Ai Hub (by Weights)

[covert lake]

gm AI HUB by Weights

[fallen plover]

I wonder if there is any way to train LoRA models locally

[gaunt shard]

Hello, I am a passionate AI / ML Developer with over 6 years of experience crafting and deploying advanced AI solutions, who specializes in Deep Learning, Natural Language Processing (NLP), and Gen AI, with a proven track record of building high-impact AI models and applications. Driven by a commitment to leveraging AI to push technological boundaries and tackle real-world challenges, eager to contribute expertise to pioneering projects at Your Company.

[sand nest]

.

[umbral bronze]

ello

[wind stratus]

with kohya, theres tutorials on youtube

[fast nacelle]

hi

[night lake]

Co-knee-chi-wah

[odd knot]

helo

why is the slowmode enabled?

[split whale]

Why it shouldn't be?

[covert lake]

why not

[magic charm]

greetings fellow meatbags

#covers #originals etc not a thing anymore?

5 seconds is not a very restrictive slowmode 🙂

[upbeat plover]

blyuv

User Avatar

[hollow kindle]

HOWDY

[solar torrent]

How what?

[hollow kindle]

can i chat with a mod

or admin

[solar torrent]

Is there anything you want from them?

[hollow kindle]

My main account "brand00dle" got hacked and was sending links in the chat resulting in ban, just wanted to see if I could get unbanned lol

[solar torrent]

You could ping one who you think they're active.

[hollow kindle]

❤️ i'll try

[feral halo]

Ayo? @hollow kindle level 1 !!!

[hollow kindle]

@little bobcat

[little bobcat]

Sup

[hollow kindle]

My main account "brand00dle" got hacked and was sending links in the chat resulting in ban, just wanted to see if I could get unbanned ❤️

[solar torrent]

How do we know if you've recovered that?

[hollow kindle]

i can message here with it when i join i guess

I will mention backshots lmfao

miss that pepe kissing sticker 😦

[solar torrent]

Rip and tear until it's done, by AI Hub by Weights.

[obsidian jungle]

Guys where can I find good Kling ai prompts?

[floral cairn]

How do I use the colab?

[polar flax]

I suck at teaching step by step, but someone has made a doc that may be friendly to who haven't used colab notebooks: https://rentry.org/msst-colab

[floral cairn]

Alr thanks!

[nocturne elm]

Human-Interest Feature

Prompt
Write a human-interest feature that highlights a problem statement using the autobiography of malcolm x by haley alex and at least two other individuals’ stories from your research.

Requirements:
This article must be written in first person (I, me, my, myself) from malcolm x
Unique (“Feature-like”) title
Submit Google Doc with editing rights via Canvas before starting
750-2,500 words, single-spaced TNR, Arial, or Lora
For citations, use author’s last name and the page number: (Orwell, 23), then cite all texts at the end of the article using MLA style in a “Works Cited” section.

Rubric
Clarity
Conciseness
Consistency
Pathos

Reminders
Begin in medias res and tell stories. Be creative, be colorful.
Only use 1st person (I, me, my, myself, we) if telling a personal story.
Avoid 2nd person (you, we).
Choose language that is impactful. Show don’t tell. Don’t say “it’s so sad” or “This problem has to be solved.” Show us why using hyper-specific language.

[night lake]

cool but like

who asked

[elder willow]

Nah Wrd also we about to nut in ur butt

[night lake]

Cool story bro 😎

[prime crow]

Hey guys ! do you guys have a link for the latest rtx version for okada voice changer ?

[solar torrent]

Himan-interest prompt deez nutz.

[pine zenith]

p

[worthy coyote]

can we have political conversation here

[slender finch]

Can someone translate

What u say 😡

[solar torrent]

[slender finch]

What u say about my mom 😡🥵

[solar torrent]

I don't understand what bro even talking about.

[slender finch]

I only understood the “what u say 😡🤬🥵

[feral halo]

Ayo? @slender finch level 1 !!!

[solar torrent]

https://tenor.com/view/moodeng-gif-17085597799575991521

Bro still speaks in Russian.

[worthy coyote]

is he drunk?

[solar torrent]

Are you?

[elder willow]

can you stfu

[worthy coyote]

@worthy coyote

[elder willow]

god damn always saying sum shi that isn’t important

Like bro GTFO

[worthy coyote]

why is bro so angry

[solar torrent]

Was I talking to you before? Why are you so serious about that?

I ain't even talk to you before bro. Before you telling me to shut up, would you like to calm your ass down also?

[worthy coyote]

i think he's drunk

[elder willow]

funny how you think I’m not calm

just shows how weak of a man you are

[solar torrent]

Sure, I know it.

[elder willow]

oh and you have autism

So explain to me how does it feel being treated differently from us neurotypical

[elder willow]

I mean it must suck being treated so condescendingly

[solar torrent]

You say it like you know me before, even if I don't even know or talk to you before.

[elder willow]

it’s easy to tell you not hiding it

[worthy coyote]

bro prepare some popcorn

bros are fighting

[solar torrent]

https://tenor.com/view/eating-the-chip-chips-chip-eating-chip-man-eating-three-chips-gif-18885184

[worthy coyote]

easiest way to end the keyboard fighting: pull up

[solar torrent]

You'd better close your Discord and go do something else in your life instead of fighting me with your nonsense argument that will give you enough attention. I can play with people as much as I want, but at the same time I can show respect to others. But since you disrespect me for the problem I didn't even make, I have no choice but to argue back with a reason.

You don't expect me to get better, boy. You're just projecting about your issue. Get over it.

[queen kernel]

Which is the best alternative of elevenLabs for TTS (local)

[solar torrent]

GPT-SoVITS, as Nick said.

[queen kernel]

Does it work offline?

[solar torrent]

I'm not sure which fork or program for AI TTS runs best.

[queen kernel]

Do it have any pre trained models

Lemme show you an example what kind of voice I want

https://youtube.com/shorts/_baOcwBW1uA?si=TagWSGTk0wLs2jRp

@solar torrent

[covert lake]

Yeah
GPT-SoVITS for few shots training
F5 or fish speech for 0shot training

U can choose one of those locally

[queen kernel]

I want this kind of voice

@covert lake

[queen kernel]

Didn't understand that.. what it is

[covert lake]

Dunno whose voice is that tho

[covert lake]

0shot training= no explicit training needed, you just gave it a short audio file of the voice and it works

[queen kernel]

I think he genrated this voice from elevenlabs

[covert lake]

Few shots = it needs actual training which takes more time and resources

[queen kernel]

Do it came with some pre-trained models which I can use ?

[covert lake]

In both cases you need to give the ai an example of the voice, maybe could work if you give it the one of the video

But you can't generate a whole new one of out nowhere

[queen kernel]

Do it came with some pre-trained models which I can use ?

[covert lake]

F5 and fish speech no

They are just audio files

No model training, u just put the audi file of the voice and it works

While gpt so vits I don't remember tbh

[queen kernel]

I tried it on hugging face space but it was too bad

[covert lake]

But you can find premade one by other users in #1175430844685484042

[covert lake]

Which

[queen kernel]

Lemme send you the link

https://huggingface.co/spaces/lj1995/GPT-SoVITS-v2

[covert lake]

Also 0shot is always inferior in terms of quality than few shots ofc

[covert lake]

That's the 0shot only

Not the few shots one

[queen kernel]

Where is the few shots

??

[bright flower]

hi

[hot sentinel]

hello

[covert lake]

i dont think there was an huggingface space for that

[smoky yew]

Is there anyone here who can separate or sing the ad-libs, main vocals, and overlapping vocals of Red Velvet's Peek-A-Boo?

[covert lake]

however the local guide should be https://rentry.co/gpt-sovits-guide

i think this https://docs.ai-hub.wtf/tts/gpt-sovits/ is only for the beta/v1 instead of gptsovits v2

[feral hemlock]

i asked a small minded question and deleted it

[polar flax]

over 3,000 of Matsushita's voice recordings
https://www.techspot.com/news/105764-panasonic-resurrects-long-dead-founder-ai-share-management.html

[solar torrent]

Their AI system will be powered by their batteries.

[polar flax]

AI robots that can charge themselves on a nearby outlet

[solar torrent]

https://media.education.studio7thailand.com/9793/Panasonic-Battery-Alkaline-AAx4-2-square_medium.jpg

They will power their AI by this thing, indeed.

[analog cosmos]

dump the batteries into the ocean

[feral hemlock]

is the voice changer down or is it not made yet

[feral halo]

Ayo? @feral hemlock level 1 !!!

[covert lake]

you replied to a message that has nothing to do with voice changer

What's ur pc gpu and what are u talking about

[feral hemlock]

its tts no

but there is a part of it that says voice changer

[vocal root]

this is the worst idea ever

[covert lake]

where?

[solar torrent]

-realtime

[rare sorrel]

Interaction has expired, use the command again for a new interaction.

💻 Local Realtime RVC

• Guide for deiteris' modified W-Okada's Voice Changer Client (Fork)
• Guide for W-Okada's RealTimeVoiceChangerClient
• How to use real time conversion with RVC (voice changer)
• RVC TTS

[solar torrent]

Huh?

[feral hemlock]

i cant really send images so i'll dm you

when you launch the webui it opens a tab, with that tab there is 3 panels, one of them says voice changer

[solar torrent]

[covert lake]

u sure ur talking about gpt-sovits

[feral hemlock]

im pretty sure

it has all the same features and the same exact settings

[covert lake]

Don't really remember it tbh

I have used it only on my own cloud notebooks

[feral hemlock]

alr

[errant sphinx]

That more powerful, 4 is used by camera digital! 🤔

[wheat star]

Hey, guys, can you give me a link to github to create a model

[covert lake]

what's ur pc gpu

[wheat star]

AMD Radeon R7 240

[covert lake]

damn, that's too weak to train (create) rvc models locally (on ur pc)

got only 2gb of vram lol

You can Train RVC on cloud (remote good pc):

1. Prepare the Dataset
2. Setup RVC:
   Choose a cloud way to use RVC,

• 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)

Google Colab = Easier but risk of getting disconnected
Kaggle = Harder but way more gpu time

3. Be sure to know about the tensorboard

If you are looking for the easiest way and for free, is using https://weights.gg which ofc uses RVC

[wheat star]

today, I used RVC-GUI For AI cover It was working fine for me.

[covert lake]

RVC-GUI is super outdated

don't follow yt tuts

also, you're using an RVC version which doesn't have AMD support, meaning it used your CPU rather than your GPU

you can inference (use models, make ai covers) on CPU but it's SLOW ASF

A minimum gpu for inference would be a GTX 1650, but yours is weaker

and training on CPU would take literally more than days, if it even works without crashing your pc

@wheat star AI takes ALOT of resources, inference less, but especially training needs more resources

Don't use RVC-GUI, I really suggest you to not use any rvc locally on your PC

You can RVC Inference on Cloud (remote good pc, easier and faster than ur PC but it's limited):

• Ilaria RVC Zero: fastest and simplest that you can get for free
• Weights.gg: Partnered with AI Hub, lets u do them easily but u may be in a queue
• Applio Colab: max 4 hours, not granted, of GPU

[kind rose]

qq all

[covert lake]

qq?

@atomic vector explain the meaning of 'qq all'

[atomic vector]

"qq all" is a phrase that originated from the world of online gaming, particularly in massively multiplayer online role-playing games (mmorpgs).

"qq" is short for "quick quit" or "quit quickly," which is a term used to describe when a player suddenly leaves a game or a group, often without warning.

when someone says "qq all," they're usually expressing frustration or disappointment that their team or group is doing poorly, and they're expecting everyone to quit or leave soon. it's like saying "we're all gonna quit anyway, so why bother?"
-# AI-generated responses may be inaccurate; please verify important information.

[ionic pumice]

qq all

[odd knot]

new roles?

also new icons for the model maker roles???

crazy

[covert lake]

🔥

[worthy coyote]

sometimes you can find out the "what the heck is this sticker for" stickers
for example, like,

[worthy coyote]

srsly what is this for

[spiral breach]

Do you guys use the voice changer to troll or to use it because it's fun?

[worthy coyote]

erm have never used it

[worthy coyote]

but they would?

[worthy coyote]

intersection of big discord servers is that they have weird emojis/stickers maybe

[dawn temple]

What is that hahaha 😭

[worthy coyote]

ai hub original

i genuinely wonder who is decided to add this sticker lol

[analog cosmos]

I didn't even notice it until now tbh

[worthy coyote]

what did i just say

[limber cliff]

https://cdn.discordapp.com/attachments/1259880961840119903/1262409159316475966/sig-4619552.1210682__safe_solo_rainbow_dash_screencap_animated_edit_castle_sweet_castle_spoiler-colon-s05e03_windows_error_message.gif

[analog cosmos]

https://tenor.com/view/jesus-ballin-mars-bars-gif-19910027

[covert lake]

ai hub best sticker

[worthy coyote]

i miss the old kanye

[river verge]

Dilly ding, dilly dong! A new RegalHyperus drum model just released!
Haaland (Ha Ha Ha) V2 (Drum model no. 546)

[elder willow]

-rt

[rare sorrel]

Interaction has expired, use the command again for a new interaction.

💻 Local Realtime RVC

• Guide for deiteris' modified W-Okada's Voice Changer Client (Fork)
• Guide for W-Okada's RealTimeVoiceChangerClient
• How to use real time conversion with RVC (voice changer)
• RVC TTS

[fallen plover]

is there any way i can run ilaria rvc offline like locally on my own gpu

[chilly lake]

get a local install fo RVC Mainline or Applio

as long as you have some decent GPU

[fallen plover]

i already have applio installed, and i have a rtx 4080 super

[kind hamlet]

👋

[plain cove]

@spark yew -> #1159290752195633273

[patent axle]

Guys i need help
Probably i sent message to wrong channel but still: #📑│making-models message
Help me please if you can

[candid harness]

👋

[ruby wharf]

e

[elder willow]

sd

[thorn cedar]

hi

[errant sphinx]

They did not work your use emote from FakeNitro User, because it can only show out links only! 😏

[ionic pumice]

https://tenor.com/view/epic-embed-fail-krookodile-krookodile-fail-gif-25018211

[sweet nymph]

What

[solar torrent]

https://tenor.com/view/cat-fall-epic-embed-fail-fail-funny-gif-21341768

[drifting ferry]

so im trying to naviage through the channels, whats the software for Ai Voice?

[solar torrent]

Real-time or an audio conversion?

[drifting ferry]

imma guess real time? since its for RP porpuses for Discord and such

[solar torrent]

-realtime

[rare sorrel]

Interaction has expired, use the command again for a new interaction.

💻 Local Realtime RVC

• Guide for deiteris' modified W-Okada's Voice Changer Client (Fork)
• Guide for W-Okada's RealTimeVoiceChangerClient
• How to use real time conversion with RVC (voice changer)
• RVC TTS

[drifting ferry]

ah i have to do the command myself

i appreciate the help

[solar torrent]

W-Okada is real-time voice conversion program, while RVC and Applio are audio conversion program.

[drifting ferry]

so W-Okada modulates and modifies ur voice real time while im guessing Applio uses recording

[feral halo]

Ayo? @drifting ferry level 1 !!!

[drifting ferry]

-realtime

[solar torrent]

Yes, that's what all about.

[rare sorrel]

Interaction has expired, use the command again for a new interaction.

💻 Local Realtime RVC

• Guide for deiteris' modified W-Okada's Voice Changer Client (Fork)
• Guide for W-Okada's RealTimeVoiceChangerClient
• How to use real time conversion with RVC (voice changer)
• RVC TTS

[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

[drifting ferry]

gonna read both guides too see which one works best

[forest crypt]

https://tenor.com/view/sound-the-car-alarm-cat-gif-9895924581726852336

[tranquil cosmos]

@hidden grotto

[hidden grotto]

:wave: @tranquil cosmos, How can I help?

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models
• /create - Create an AI Cover
• /image - Generate an Image

[dreamy turret]

Helloooo!

[night lake]

huh when did 11labs get an ai speech detector

huh didnt know ai hub was in a research paper

[solar torrent]

AI Hub (by Weights) mentioned.

[covert lake]

Wow

[covert lake]

Ai detectors aren't really it tbh

Idk about speech, but for LLMs it's really easy to make it sound human

[polar flax]

only for 11labs outputs, not the one using rvc?

[lethal elk]

hello everyone i am Luna.....AI is my life...thank u

[odd knot]

yeah maybe

there are also 11labs ai chatbots iirc

they're only 4 tho

[covert lake]

AI 🔥

[covert lake]

damn since when

[odd knot]

idk

like, 2-4 months ago?

[covert lake]

ur new role icon looks better now

it was a butterfly before lol

[odd knot]

you mean a black box with the "qc" text?

[covert lake]

no before this orange one, it was changed to a pink butterfly

you're talking about the og one instead

[odd knot]

oh

i didn't really notice that

[polar flax]

https://tenor.com/view/露西亞-gif-26180620

[covert lake]

who's that character

[odd knot]

uruha rushia

[covert lake]

and it was a pink butterfly smh

well this shows more that it's a helper icon

[covert lake]

never heard of it ngl

[odd knot]

she's a vtuber that graduated back in 2021-2022

iirc

[covert lake]

oh yeah idk vtubers

[polar flax]

this was even uglier
https://cdn.discordapp.com/attachments/1159290096458149938/1312432162783629342/image.png?ex=674d21f3&is=674bd073&hm=35b78ad8d6406e4770b9715576654cb66c905f859ec1f926925dd330122b8796&

[covert lake]

the issue is that the icon is small so u need a simple one

the one we have now is more uunderstandable as it's simpler

[covert lake]

https://www.youtube.com/watch?v=dJPmj0LrEJs

[analog cosmos]

[polar flax]

I knew her till she was gone from hololive

||and also the present her as another persona||

[covert lake]

I don't really know the hololive

[odd knot]

i've heard she got married with mafumafu and the drama between those two just spreaded on twitter

[dusty glacier]

You mean 𝕏??

[solar torrent]

Japanese Amber Heard

[snow echo]

https://x.com/venturetwins/status/1862910201113739328

[covert lake]

tf that's wild

[soft arch]

help, I got No module named 'gradio'
and I don't know how to fix it

[snow echo]

AI or human, free speech is a central pillar of learning

if anything, that actually makes sense

[covert lake]

What are you using

Be sure to NOT follow yt tuts

[snow echo]

censorship has almost never had a net benefit in human society in general

[covert lake]

ur name confused with razer for a sec @night lake lol

we got Razor & Razer 🔥

[soft arch]

applio

[feral halo]

Ayo? @soft arch level 1 !!!

[covert lake]

elaborate more

What's your PC gpu and what are you looking for

[snow echo]

Haha

[feral halo]

Ayo? @snow echo level 1 !!!

[soft arch]

amd rx

6650 ti

[snow echo]

yay less go, now I can finally upload a voice model in here

I thought the req for Lvl1 would take some time but hey, that was nice

[covert lake]

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

For issues ask in #✨│ai-help

this isn't an help channel

[soft arch]

ooo thansk

[covert lake]

Not really

You need to be model maker

#outdated-model-maker-role

[snow echo]

Oh, ok

[covert lake]

With lvl1, u can send images in help channels,
lvl5 = embed perms everywhere

[snow echo]

Ah gotcha

damn, why does embed req Lvl5

[analog cosmos]

[snow echo]

just filled it out

[covert lake]

You have made a model already?

[snow echo]

I had already uploaded it to weights.gg yesterday

it has been some time since I made it

[covert lake]

oh lol

[pearl ember]

yo

[polar flax]

https://tenor.com/view/ai-sponge-embed-perms-gif-2722780943144073359

[snow echo]

W H Y

why

[snow echo]

ngl this stuff is pretty cool

[covert lake]

yo

[snow echo]

I swear man, developments so fast, it feels like a whole new playing field if you drop out the scene for a couple months even

[covert lake]

did you develop anything other before?

like are u a dev or nah

[strange orbit]

greetings fellow guys and gals!

[snow echo]

Not really no. Voice model specifically this is my first one, which I got into after trying out the W Okada Voice Changer.
Before that, it was some Machine Translation stuff for a Japanese VN and image generation after Stable Diffusion hit the scene, mostly early 2023

--
since then, I have only been keeping tabs on a very surface level fashion

[covert lake]

Machine Translation stuff for a Japanese VN
Like using an LLM to translate your visual novel game?

image generation after Stable Diffusion hit the scene
Now flux.1-dev beat sd in terms of quality, but it's more resource demanding

AI develops really fast ye

[snow echo]

The software was a model called Sugoi Translator Toolkit (in conjuction with Translator++)

[feral halo]

Ayo? @snow echo level 2 !!!

[snow echo]

noice

[covert lake]

Oh yeah idk about it

[snow echo]

ah really, so flux is the deal now?

[polar flax]

if you have 3060 12 gb you can try flux locally

[snow echo]

I got a 4060 Ti 8G

[covert lake]

yep, best quality rn

[snow echo]

not enough VRAM I know

[covert lake]

literally dominated open source and closed source image gen models

[polar flax]

or get the 16 gb variant

[covert lake]

Yea not enough

flux.1-dev needs ATLEAST 12gb iirc

flux.1-dev is the best open source one

there's also flux.1-schnell but it's like the 'lite' version

[snow echo]

So is it the way to go for everything like uhmm...high culture stuff as well

[covert lake]

high culture stuff?

[snow echo]

uncen

or is it handicapped like newer SD models

[covert lake]

I don't really use AI for that but it's uncensored iirc

[snow echo]

Ooh, imma try it out then

[covert lake]

sd3 failed bc they censored it that much that it couldn't even do normal things

sd3.5 is better tho ig, but not as flux

[polar flax]

flux dev has highest quality metric, but SD 3.5 may have best prompt adherence

[covert lake]

i think they even made a lora for sdxl trained on horribly sd3 images

[snow echo]

https://x.com/zebulgar/status/1850230824974622874
relevant

[snow echo]

yeah, that was a shitshow

funny how censorship does the same thing to brains, AI or human

[covert lake]

only the large one has better prompt adherence

I'd take sd3.5 large as a better flux.1-schnell imo

as it's a bit better than it and less resource demanding

[snow echo]

I see

[polar flax]

flux dev vs SD 3.5 in a mugman prompt test #🏙│ai-images message

[covert lake]

the 1st image is made by flux right?

[polar flax]

yep

[covert lake]

imo that one looks better

the 2nd one is kind off uncanny

literally isn't even attached to the body 😭

[polar flax]

but the prompt adherence maybe

[snow echo]

aight so last I remember we only had Stable Diffusion as the main open source player that you could run locally.
Now there's Flux but also stuff like PixArt and et al.

so are there other ones as well which I should be aware of?

[covert lake]

I mostly use Open Source AIs tbh

[snow echo]

night and day difference in that particular instance

I am not against using proprietary stuff as long as it delivers and gets the job done but open source usually comes with more mod/tinkering potential

[polar flax]

too bad I didnt refer to the prompt used 💀

[covert lake]

For closed source LLMs I rather use ChatGPT or Gemini 1.5 pro (free with aistudio)
For open source, Llama

I can run llama 3.2 3b even on my laptop & phone

[covert lake]

yea 😭

if u say mugman, i'm gonna think of the 1st image u sent

[polar flax]

there is also Qwen 2.5 72B instruct in huggingface space

[covert lake]

I thought that was mostly for japanese

[covert lake]

smh never followed qwen

[worthy coyote]

nah nah nah that that don't kill me

[solar torrent]

They only make me stronger.

[plucky halo]

Guys sent me a link to yall vocal remover Ai plsssss

[ionic pumice]

x-minus.pro, mvsep.com

use bsroformer for seperating the vocs/inst, then uvr de-echo for removing echo

[solar torrent]

You can run demucs on Colab to seperate track stems from an audio.

[ionic pumice]

yea that too

[faint cape]

hi

[static isle]

hi

[formal hare]

Guys, what should I do if I experience a large delay in my voice changer? When I test it in the program, delay is about 3-4 seconds, but when I join the game, it shows around 30 000 ms. Please Help(((

[worthy coyote]

quit it

[feral halo]

Ayo? @worthy coyote level 12 !!!

[worthy coyote]

best way

[ionic pumice]

diffsinger making peak

lol no idea why it's quiet

[polar flax]

mvsep.com, x-minus, or jarredou's/my tweaked version of MSST colab with some best mel roformer models: #1159290752195633273 message

[plucky halo]

How do u use colab vocal remover

Thxxxx guysssss

[pallid lion]

whats like the best ai voice for like an american latina girl that sounds realistic cause theres too many

[worthy coyote]

shout out to solo

who become the part of the ai hub sticker

[short dragon]

👋

[worthy coyote]

no way...

[solar torrent]

https://tenor.com/view/azumanga-daiho-azumanga-osaka-ayumu-kasuga-gif-14774670791157440244

[tired jasper]

more than 5 days training F5 for portuguese 😮‍💨

still not ready

[worthy coyote]

ready or not

[tired jasper]

increased learning rate now to see if it finish today

are you brazilian?

the clone is perfect, but spelling still not ready

i'm experiencing the "almost" for long time

maybe i should set larger learning rate since beginning

i used the f5 portuguese in huggingface as start checkpoint, but there was a huge english accent

i'm finetuning it for brazilian

f5 is far the best TTS model yet

[worthy coyote]

give up!

you'll be comfortable

[minor blade]

There's no such thing as a "best girl model"

It's matter of looking at the #1175430844685484042 channel and testing with various models to see if one fits your voice.

[covert lake]

what

[elder willow]

bro no voice is stable for me idk i have good internet and gpu is it just the configuration or smth wrong with my voice

[covert lake]

I'm guessing you're talking about Wokada, try to elaborate more in an help channel -> #🔍│help-w-okada

[elder willow]

oh correct

[ionic pumice]

affirmative

[elder willow]

how to make text to speech?

[elder cradle]

Hello

[obtuse robin]

https://garticphone.com/ru/?c=03f1506594

[whole fossil]

Hello

[spiral aurora]

5days?

[tired jasper]

yeah, totally new language

[tiny cloud]

hey im just here to chill

how do i access which level im on btw?

[analog cosmos]

same

[analog cosmos]

do /rank in #🤖│bots should show you what level you are

[vernal canopy]

What exactly does fairseq do in RVC

[lunar glacier]

.\

[lethal flax]

is there an ai for zooming out/uncropping on videos

I know that gen 3 has one but it costs money which is a thing I currently don't have

[zenith fiber]

hi

[sleek oxide]

Hello, i am new, hoy can i get a sample of cillian murphy voice?

[ionic owl]

Weres the chatt bar

[forest crypt]

https://tenor.com/view/sound-the-car-alarm-cat-gif-9895924581726852336

[ionic owl]

Weres the cattt bar

[forest crypt]

https://cdn.discordapp.com/attachments/1049644155736969267/1285635867532328991/rapidsave.com_v3hri3kg41od1.gif

[drowsy kernel]

how do I get inanimate insanity mepad voice model

[solar torrent]

#1175430844685484042

[drowsy kernel]

there is none named mepad

there's no model

[polar flax]

learn how to make models here: https://docs.ai-hub.wtf/essentials/how-to-make-voice-models/

[solar torrent]

If a voice model you wanted doesn't exists, you can train one locally or on cloud service, or request someone to do it for you at #1159289738314919936

[worthy coyote]

finally

cat + bird

[azure valley]

Hello

[wraith quartz]

Hello

[worthy coyote]

Hello

[coarse fulcrum]

HI :3

[lavish latch]

hii

[marble wolf]

hi

[covert lake]

#1159290752195633273

[narrow mountain]

is there a rvc successor or when is rvc 3 being released? plz no F5 or any other tts only voice to voice open source

[solar torrent]

Hi.

[covert lake]

Nope

There's no RVC v3

[solar torrent]

Currently, there's no RVC v3. It's all just unofficial fork of RVC v2.

[covert lake]

The fork ain't released yet unfortunately

The best program since almost 2 years is just RVC v2

[fossil maple]

Anyone know any good girl voices?

[solar torrent]

I don't know which voice model for girl or woman is the best. But voice models here were trained on famous people.

[river verge]

Dilly ding, dilly dong! A new RegalHyperus drum model just released!
The Last Element (Drum model no. 547)

[pastel mauve]

Hello! We're live on Product Hunt! 🎉🎉🎉

We transcribe and summarize audio files and Youtube videos.
It is very important for me to upvote, thank you in advance.
I also want to give everyone who comments a code to try it for free. for 1 month!

https://www.producthunt.com/posts/voiser-ai

[solar torrent]

#1159290752195633273

[worthy coyote]

助けてくれないか

え、なず俺日本語を

[wooden crown]

any FLUX lovers here?

[feral halo]

Ayo? @potent summit level 1 !!!

[cerulean coral]

Hi all, new here 🙂
Was wondering whats the best model I can use to edit videos I have, like add a hat to someone and maybe some effects..
I use alot of chatgpt and image generation, but I am a newbie in videos 😅

[queen kernel]

we work with Voice to voice conversion. But you can Ask in #🔍│help-ai-art

[cerulean coral]

oh okay, thanks

apologies, its sometimes hard to understand the purpose of each room 😅

[queen kernel]

dont worry.

[queen kernel]

you can use adobe firefly

[tulip hornet]

Halo is there any better realtime voice changer than w-okada or there's none

which I can host locally*

[quartz roost]

You can't use famous people's voice for monetization or commercial use.

[covert lake]

Wokada is the best

[open iron]

hold on

oh alr i though i lost my levels

[maiden schooner]

Is the google shit fixed now or does your account still get banned for using it?

The easy gui i mean

[covert lake]

Google Colab doesn't allow any web uis in the free tier

But now we encrypt the code, so you would almost never get banned as it doesn't get detected

[covert lake]

EasyGUI is a fork (modified version) of RVC

btw what's ur pc gpu and what are u looking for

[cobalt badge]

hi.

[tepid basin]

Ai hub

[night lake]

by Weights

[tepid basin]

My apologies

[solar torrent]

When did I say you could use them for commercial purposes? There's no voice model can be used for that. What are you talking about?

[nocturne trail]

Should I use a pretrain to get the best quality?

Barely any idea what a pretrain is

[tepid basin]

generally recommended to use OG pretrain

Custom pretrains have some issues, still being researched. Most models should use the default pretrain

[nocturne trail]

So they just make it faster and better?

[tepid basin]

Not faster

Ideally better, but there's issues with custom pretrains quality

[nocturne trail]

okay so if I want to clone a voice I should get a dataset and a pretrain?

[tepid basin]

There's some lines on the spectogram that doesn't happen with OG pretrain

[tepid basin]

Don't worry about the pretrain, just use the default one in RVC

But yeah you need dataset

[night lake]

noobies fixed that i think

[nocturne trail]

oh it has a pretrain?

[tepid basin]

Real?

[night lake]

needs to be merged tho

[nocturne trail]

you can tell i have no idea what im talking about

[cobalt badge]

Hot take maybe ai is fire fr.

[tepid basin]

Ye its built in

[nocturne trail]

ah i see, just follow the normal instructions then?

[tepid basin]

You just need your dataset

Ye

[nocturne trail]

alr thanks

[tepid basin]

Can I merge it

[cobalt badge]

does this server also include deepfakes?

[tepid basin]

Lol barely

We are mostly ai voice changer server but you might find some smart people who have done deepfakes before

[nocturne trail]

oh also what is the best way to stop being disconnected

[tepid basin]

To not use colab

[night lake]

oh nvm it got merged

[nocturne trail]

oh lol

[tepid basin]

Can I use it

[night lake]

just download the newest version of applio

[nocturne trail]

i thought there was a workaround

[tepid basin]

FUck yeah

[nocturne trail]

i have a 2060 laptop and uh i wanna do other things

[tepid basin]

Uh there might be, but i don't use colab

[tepid basin]

Fair enough. It might be worth trying on the 2060, train it overnight, but colab is prob faster

Open any game alongside it and it dies

[nocturne trail]

yeah exactly

nah the laptop couldnt handle it and chrome lol

[polar flax]

the alternative is to use a zero shot TTS, e.g. fishspeech, so you dont need to burn your laptop to train an rvc model

[errant sphinx]

https://tenor.com/view/gently-open-the-door-robot-leaves-me-no-choice-gif-14368586

And see out for hanging it!!

[wide cargo]

why is everyone by weights

[gray rover]

the " fix for lines " is tied up to mrf from what I recall, btw

it won't work that way with current pretrains sadly

( Unless we talk about something else )

[chilly lake]

The new loss function does cancel the uspcaler lines rather quickly

with the regular generator/discriminator

clearing the topmost range is a bit problematic still

perhaps need to tweak the clamping value

[gray rover]

oh then a different thing is talked about

either way, what loss function?

[chilly lake]

mrf itself does not fix those upscaler lines, but it does fill the gaps

MultiScaleMelSpectrogramLoss

[gray rover]

oh, that

aaaaa, that was the context lol

nevermind. Thought you added something else in there

[chilly lake]

nah, but it was a minimal fix that does something good without much overhaul

[gray rover]

off topic, Imma give 48 more trials