🧬│ai-chat

[edgy kiln]

This video was made with Google Gemini https://www.youtube.com/watch?v=Ql80R5danzQ&ab_channel=Славик0001

[solar torrent]

This is not where you request anyone here to train a voice model for you, and that request channel has been removed.

Read the latest on #✨│announcements for more on why.

[oak bluff]

Hi guys
I have a chemistry assignment that is to make a parody about the subject
I wrote the lyrics thinking about Soad's Vicinity of Obscenity and I need a voice that at least fits the rhythm of the song
Can anyone help me?

[soft hornet]

Why can't you sing? Btw I'm chemistry student

[soft hornet]

Use SunoAi

[chilly lake]

try ace-step project, it is free

[vast sky]

Hey guys... I made a new kind of music with AI... based on semantic destruction and the preservation of "human soul" in the structure. It feels like music from another dimension. The album is already on the way. This is SimCore.

[copper quiver]

hi , AI chat

im looking for some files
any help !

[oak bluff]

Initially I was going to do it with Ai, as its use was authorized and I believe it would be better, but I was thinking some time ago about singing it myself

[oak bluff]

I tried a few times but it wasn't exactly what I expected

[oak bluff]

I will try, thank you

[tacit widget]

Looking for a good TTS that has an api and takes in ofc a pth and index. Suggestions?

Local

[chilly lake]

you can run a standalone tts, then you can run a voice change

but you can also just run a tts with zero shot voice cloning

(available options depend on the language you need)

[turbid vector]

Reaal

[quaint path]

hey guys, im new here , someone can tell me a program to use real time ai voice changer?

[copper quiver]

i need help

[elder willow]

anyone here know how to use the models for w-okada (voice changer)

[plush sail]

yo guys whats epoches

[chilly lake]

epoch = a loop thru the entire training dataset during training

[plush sail]

how do i train epoches

i hear people saying its good to make sound clear

[chilly lake]

you train a model, an epoch number is just the number of loops what the model has done thru the entire dataset during training

[plush sail]

where do i train

pls teach me how to do epoch training

[chilly lake]

https://ai-hub-docs.vercel.app/essentials/how-to-make-voice-models/

[plush sail]

but i downloaded a index and pth is that it or what

[chilly lake]

and index and a model's pth is a result of training

[plush sail]

if i downloaded a custom one in voice models it say trained 30 mins but how do i train bc i hear that 800 epoches is good how do i do that

[plush sail]

what dose it mean here they trained epoches dosent the thing already have its own https://discord.com/channels/1159260121998827560/1175430844685484042

[chilly lake]

it just means how long those models were trained

[devout sky]

IDK IF ANYONE LIKES YE BUT IF U DO

TAP IN 10:00

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

[odd glacier]

hey im new here why cant i hear the voice changer on discord

[hidden bolt]

Hi all, i have been tinkering around with the realism section. I can't seem to get Light Host to launch

Does anyone knows why or how to fix said issue?

[finite meteor]

which online realtime vc that's still works?

at least uh, for amds

[plush sail]

so if i train it on my own would it be better or nah

[chilly lake]

if you've never done it, most likely it wont be good at all

it takes some practice to make something actually good

[plush sail]

do u have any good models that are a egirl sound

i need to be a egirl

[azure zodiac]

Hi everyone Im the creator of Planet Warz, i have been solo developing this universe for the past 4 years, working on a book series, a video game and i have created all the concept art, music, and managing the social media aspect of all this. if someone need any tips, feel free to hit me up

[tender canyon]

how do i use an ai model?

[floral cairn]

Has anyone read AI 2027? Really interestimg to see how fast AI is going and we are living through this moment in human history

[paper flare]

is it documented how does the character of the voice affect its quality? or its not applicable

for example, do nasal voices have better quality than others and/or require less epochs to train?

[frank bough]

Is there any working colab to train RVC ?

[nimble ether]

ih hub ai not free?

[supple stream]

what

[nimble ether]

eeh

voice ai

[nimble ether]

its not free?

[supple stream]

What is not free?

[frank bough]

anyone? 😦

[nimble ether]

yea not free like if i open the app pop up paid is on idk sorry for bad eeng

[ashen yarrow]

Where is Epochs?

[unique gale]

can the voice models be used as a live voice changer?

[prime tangle]

@neon tendon

[dawn copper]

how many minutes of speaking is reccomended for a decent sounding voice clone?

[echo inlet]

are you like looking for a live call voice changer or something for gaming like streaming?

[minor blade]

I would suggest between 40-50 mins or 1 hour of clean audio. (depending on the purpose you'll give to the model)

Also, for further help regarding model making use the #✨│ai-help channel.

[chilly lake]

10-60 min, not more

[minor blade]

Facts.

[chilly lake]

depends on the actual audio content

[minor blade]

X2

[echo inlet]

agree to this, focuses on the quality of the audio

[dawn copper]

thanks guys

[chilly lake]

like if you got someone reading tongue twisters, maybe 10-15 minutes is enough, if you got someone's podcast ranting about a footbal game maybe more to get more unique content

[faint laurel]

What are colab guys using actually?

[light copper]

how do i make the ai voice sound more believable? for trolling purposes ofc

[supple stream]

probably

this

-colab

[pine acorn]

📒 Google Colab Notebooks

Google Colab is a Cloud (Remote Good PC) Service. While the Free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

• **Applio**

by IA Hispano
Google Colab

• **RVC Mainline**

by Hina
Google Colab

• **UVR5 NO UI**

by Eddy
Google Colab

• **UVR5 UI**

by Eddy
Google Colab

• **Wokada Deiteris Fork**

by Deiteris & Hina
Google Colab

• **Hina's Modified Original Wokada**

Google Colab

• **RVC-AI-Cover-Maker-WebUI**

by Shiro & Eddy
Google Colab

• **FaceFusion UI**

by Nick088
Google Colab

• **FaceFusion NO UI**

by Nick088
Google Colab

• **Music Source Separation Training (Inference)**

by Jarredou & Makidanye
Google Colab

[tender siren]

AttributeError: 'Pipeline' object has no attribute 'get_f0_crepe_computation'

                f0_method = gr.Radio(
                    label=i18n("Pitch extraction algorithm"),
                    info=i18n(
                        "Pitch extraction algorithm to use for the audio conversion. The default algorithm is rmvpe, which is recommended for most cases."
                    ),
                    choices=[
                        "crepe",
                        "crepe-tiny",
                        "rmvpe",
                        "fcpe",
                        "hybrid[rmvpe+fcpe]",
                        "hybrid[crepe+fcpe]",
                        "hybrid[crepe+rmvpe]",
                        "hybrid[crepe+rmvpe+fcpe]",
                    ],
                    value="rmvpe",
                    interactive=True,
                )

it is intended right

[heady osprey]

guys im having this err

[chilly lake]

depends on what you're trying to use

if you're trying to use applio's exp/f0_spin, then hybrid options are no longer available

[jaunty siren]

Türk yokmu

[olive coyote]

is this the main chat?

[spring loom]

For ai talk yes

For general talk #✦│chat

[plush sail]

bro how do i fix the voice ai sounnd so laggy

[spring loom]

You're gonna have to give mroe info than that

[minor blade]

Use the #✨│ai-help channel for help regarding models or realtime

[tender siren]

i just noticed leftovers in code

[still sigil]

@strong sparrow no advertising

[strange crow]

i forgot what is the best ai cover maker rn?

[hazy lotus]

who got good voice model egirl with which you can laugh and shout without a bug? go dm plz

[hazy lotus]

yess

[austere ginkgo]

I was thinking of a new architecture i could implement

anyone knows how to actually go about the coding?

idk much about coding in pytorch so can ya'll recommend some ai that i could use

[polar flax]

what kind of architecture do you want to propose?

there are already LLMs capable for coding, such as Claude, Grok, and Gemini 2.5 pro

there are also some open source ones that can be run locally on average consumer pc

[austere ginkgo]

it's xceptionNet related , like for image detection on a deepfake model

[polar flax]

what? so "code" detection?

there's nothing could be done

but to do rigorous testing and reviews

[austere ginkgo]

to see through a video and to output weather the video is real or fake

[polar flax]

I don't get it, you said about "coding"

then you meant the deepfake videos

try addressing like artifacts or any deepfake flaws

then you'd have to feed with massive amount of detection data

[austere ginkgo]

I meant

idk how to code

to implement the model

[dapper drum]

Also, is it remove vocals then use the karaoke then use the De echo and de reverb model

[fresh coral]

i want to open a YT channel i have made a profile pic for my YT
i want to use an AI website that will generate me a banner that goes wiht my profile pic

which AI website should i use i tired midjourney they are asking me take the paid subscription

[supple stream]

Google fx

Google Ai studio

[jagged seal]

@sick cave have you made any new models :3

[frank bough]

is it normal that applio training takes 1 hour to get to 40epochs on a 33 mins dataset? When i used RVC2 it was faster.

1 hour 12 minutes actually

[white parcel]

what's your gpu and training batch size?

[frank bough]

@white parcel batch 8

T4

[frank bough]

62 epochs at 1 h 52 m 😦

[chilly lake]

T4 is about 3050 speed, shoud be about 3-4 min per epoch on 45 min set.

@frank bough

1.5min/epoch seems way too fast.

[mental bough]

why did litsa rejoin

[polar flax]

who cares

[frank bough]

when i train RVC2 instead of Applio i would get like 150 epochs on a 30 mins dataset in an hour...

i am at 130 now and it's training since 4 hours

[chilly lake]

4070 TI S, 35min set, batch 8, 130 epochs/hour

so if you somehow is getting same results with T4, which is like 5x slower, then there's something not right

[frank bough]

Tested the model sounds great so far but it’s training soooo slow

[chilly lake]

old RVC is using FP16, so it may be faster on T4

but there's a high chance of the model exploding

[frank bough]

I wanna use RVC but there is no working collab and kaggle doesn’t work at all.. only errors

[spring glen]

can someone help me install vc

[spring glen]

?????

[chilly lake]

what is there to install? download, unzip, run

[spring glen]

im lil confused, can you join vc

[chilly lake]

https://rentry.co/forkvoicechangerguide

[real pulsar]

Best pc to for live mode Ai voice changer that’s not to delayed

[elder willow]

-hugging

-huggingface

[pine acorn]

🤗 Huggingface Spaces

• **UVR5 UI**

Huggingface Space by Eddy and Ilaria

• **Ilaria RVC Zero**

Huggingface Space by thestingerx

• **RVC⚡ZERO**

Huggingface Space by r3gm

• **Applio**

Huggingface Space by IA Hispano

• FaceFusion

HuggingFace Space by Nick088

[chilly lake]

if you want to play some heavy 3D games and use a realtime voice changer at the same time, probably 4000-series nvidia at least

[real pulsar]

Not delayed correct?

[chilly lake]

very mimimal delay, the issue is that there's no concurrent schduling to 3D and AI tasks, so each one tries to take all the resources.

[echo inlet]

hi, what's your gpu? applio is actually quite slower as they focus more on the quality over speed. but yeah it's quite normal

[frank bough]

T4

[echo inlet]

T4 is solid, but not the fastest for training-heavy tasks. if applio's running with the default settings, then the pace sounds about right

[frank bough]

damn.. i need to find a RVC2 colab that actually is not banned 😄

[loud frost]

salve

voltei depois de muito tempo

[copper flume]

yo i need help with the voice changer AI bc i cant get out the pay option i have to pay 100%?

[minor blade]

Are you using some sort of sketchy voice changer?

[elder willow]

Did I really get in trouble for saying the n word

[minor blade]

Yes.

Racial slurs are prohibited on the server.

[hasty jacinth]

Does it sound like legit like fr

[loud frost]

alguem curte br ?

[chilly lake]

you mean does the voice changer sound like a real person? barely and pretty easy to figure out someone is using... unless the person is trained to use it properly and does not whisper/laugh or raise voice.

[copper quiver]

@orchid slate DM ! just if u can feed me for sone info about AI

[rich hamlet]

to

need voice changer help

plz

plz

[elfin crypt]

hey, does anyone know realtime voice changer that can run even when using intel uhd graphic?

[chilly lake]

"Intel UHD Graphics do NOT work at this point in time. Use Online Alternative."

[hybrid temple]

why does the old okada voice changer sunds better than the new one

[daring mulch]

Hi guys, I'd like to discuss an idea regarding AI Safety that I had. Anyone open to a discussion?

Currently, most alignment approaches to agentic AI safety involve trying to align the model by assigning a negative reward to an outcome/plan that is deemed harmful. But to me, that's not a good way to approach safety.

Doing this will create an even more enormously sophisticated reward landscape and does not have sufficient safeguards against misaligned AI

My idea: Separate the planning phase of an agentic AI into two modules:

1. Brainstormer Module
2. Categorical Imperative Module

The Brainstormer Module will try to find the optimal way to solve a problem and achieve the goal, and condense this information into a plan. The Categorical Imperative Module will try to reason about the plan and use cause and effect reasoning to understand the effects that it will have in the real world and its implications. Then it will check against a set of hard-coded rules ("directives" if you will) and if any of the directives are violated, then the plan is vetoed. If the plan is vetoed, then the Brainstormer has to come up with another plan, and the process iterates until the CIM receives a plan that doesn't violate any of the directives. So we can make the Prime Directive of the CIM "Do not wipe out the human race" and hopefully this architecture could help secure our safety as a species.

Thoughts?

Make the categorical imperative module completely untouchable to the Brainstormer so that it can't circumvent it, too

[elfin crypt]

well, that's what i'm asking, no? i've been searching for online alternative that can work even with Intel UHD, but most of them are just outdated or just had outdated codes that don't work on google collab. i've tried using beatrice and tweaking it around and it worked, but it still didn't work the same way prime okada did months or even a year ago, lol

i tried several ways to reverse engineer the hina version in google collab, the code worked at first, but somehow it still don't want to install torch. and i was confused, lol.

[chilly lake]

nobody cared enough to update the online voice changer and I think google banned a few

[elfin crypt]

or perhaps it's already installed but can't find the module. but anyway, i failed. miserably. and my laptop felt like it's almost exploded from trying to reverse engineer it.

fudged intel UHD for eternity

[grim locust]

do anyone know if zonos can be modified to not reprocess the voice everytime you want to generate for that voice. i meant to make it shorter the process since u process the voice already the first time

for zero shot

[chilly lake]

zonos is out, chatterbox is in

but you can just save the speaker```wav, sampling_rate = torchaudio.load(r"david.wav")
speaker = model.make_speaker_embedding(wav, sampling_rate)

[grim locust]

what do u mean zonos is out. is there no longer a zonos?

[chilly lake]

i mean quality-wise

[grim locust]

i see thanks. i dont know about chatter box. is it also zero shot?

[chilly lake]

yep

there are some minor issue with noise at the end of sentences, but it is so good

[grim locust]

thank u so much noobies ill try to look it up. this meant alot to me, i hope i can ask u questions later on if i got some

[grim locust]

for this sampling rate is the one generated from the first cloning?

does speaker mean the one that is generated from the first zero shot cloning?

[chilly lake]

one thing about zonos is that it makes 44k files

the script above loads the audio for the voice, then it passes its sampling rate to the make speaker embedding function

I think it just needs it to adjust some parameters/resample it properly

so the result of make_speaker_embedding is a tensor you can reuse if you want to infer more text to speech

or you can save it and load from file

[grim locust]

so what i need is the embedding that comes from the first voice cloning. then ill load that embedding to ur code so i wont have to reprocess the whole zero shot again?

[chilly lake]

not sure what you mean

[grim locust]

from what i understand. zero shot cloning process the voice to clone everytime u want to generate a cloned voice

i thought i could get a embedding file or something from the first voice cloning. and use it later on so it wont reprocess the whole thing again and just change the voice from that embedding

[chilly lake]

speaker is the voice parameters

hold on, lemme refresh my script

[grim locust]

im sorry im dumb at this. a have basic programming knowledge but i havent tried it on this type of things yet

ill try to install chatterbox first before asking further questions on how to make it faster.

by running the script instead of the default GUI

also thanks alot

[still sigil]

you can make it yourself on weights

[still sigil]

you can make it yourself on weights

[grim locust]

any applio experts here? i just found the embedder model option when generating from TTS. can we use our own embedder model and how do we create one?

[chilly lake]

chatterbox is not very fast, but so good...

somehow manages to change the way some words pronounced based on the speaker

[sleek chasm]

https://tenor.com/view/gotta-blast-operate-mascot-get-this-work-gif-17924681

[grim locust]

im having some problems installing chatterbox maybe because my pc is old

[grim locust]

do u also use applio?

[chilly lake]

yes, tts there is not good, just a basic thing as a demo of what can be done

[grim locust]

im trying to installed chatterbox but im not so sure if upgrading my phython to 3.10 will also affect some other programs i use

[chilly lake]

what do you have?

generally you can have as many versions as you want

you can create a specific environment using py -3.10 -m venv venv

[grim locust]

i see, i have 3.8 i will install 3.10 now for chatterbox and create venv using it before trying to install chatterbox again

hope no more problems show

[chilly lake]

3.8 is so old

[grim locust]

yes XD

just like my pc

[polar flax]

even the og rvc uses 3.9

[chilly lake]

outdated af

[polar flax]

ye mainline was using it, then ilaria rvc and applio upgrade it to 3.10

[chilly lake]

mostly because fairseq has no 3.10+ wheel

[grim locust]

yeah i upgrade to 3.10 if it fails again ill go 3.12

yo noobies, i just want to clear something.

the code u gave if i understand correctly

where do u get the speaker embedding?

[chilly lake]

for zonos?

[grim locust]

zonos or chatterbox i guess.

the speaker part

[chilly lake]

I could not make my command line work, something for screwy in the zonos build

chatterbox seems to run the speaker embedding extraction every time you run inference

[grim locust]

i meant this wav, sampling_rate = torchaudio.load(r"david.wav")
speaker = model.make_speaker_embedding(wav, sampling_rate)

[chilly lake]

that's zonos

[grim locust]

ohhh

u mean chatterbox also runs speaker embedding everytime u generate

is there a way to get the speaker embedding from the first time and just load it the second time so u dont have to reprocess the whole thing

[chilly lake]

this way it re-uses the embedding

[grim locust]

doesnt this #calculate speaker embedding
speaker = model.make_speaker_embedding(wav, sampling_rate) recalculate it again? sorry im a bit dumb

[chilly lake]

it only runs one

before the loop in the second example

but yes, if you re-run the script again it would re-calculate it again

you can do torch.save(speaker, "speaker.pth") to save the calculated value

[grim locust]

i see, so u created the scrip assuming im using it from the GUI?

[chilly lake]

then instead of loadin the .wav you can load the speaker.pth

i'm not using gui

[grim locust]

can i also do this on chatterbox?

i meant change its code

btw thank u very much for the help. i will try it after i finished installing chatterbox

i deleted my zonos before so cant try it yet.

[chilly lake]

for chatterbox there may be some changes needed to decouple the speaker calculation the the generation.

I think you can just call model.prepare_conditionals(audio_prompt_path), and then the generate without

seems to work

[grim locust]

how do u do the 2nd generation with this code? the first one i run the script how about the second

sorry if im being clueless. i just thought rerunning the script will have to reprocess it again but how do we generate the 2nd time if we dont rerun the script

[chilly lake]

I'm only concerned about delays between individual lines.

but if you really need to skip the conditional calculation every single time, perhaps you can save model.conds to a file and load it back

[grim locust]

this is what i really want to know #calculate speaker embedding
speaker = model.make_speaker_embedding(wav, sampling_rate). is there a way to hard code this like putting the previous result file or something

so it wont recalculate speaker embedding*

[chilly lake]

I've explained it already

#🧬│ai-chat message

[grim locust]

torch.save(speaker, "speaker.pth") so i will use the .pth that is saved from first

thanks

last question what would the code look like when i use that .pth

[chilly lake]

it will save the embedding into a file

to use it you just do speaker = torch.load("speaker.pth", map_location=device, weights_only=True)

[grim locust]

this is it

its clear now

please dont be mad on me, i just realize chatterbox doesnt have this speaker thing on the code

[chilly lake]

please focus on one thing, so I dont have to mix two project's explanations.

I've explained how to do it with chtterbox too

[grim locust]

correct me if im wrong but its here right? wav_fpath=AUDIO_PROMPT_PATH

this is the one i will replace with .pth

[frank bough]

yo, anyone know a noUI colab for RVC2? not applio pls.

rvcdisconnected not working anymore

[grim locust]

really sorry noobies, its on me for being confusing. i thought they have same python code

[chilly lake]

model.prepare_conditionals(wav_fpath=AUDIO_PROMPT_PATH) call creates model.cond value, you save it to the file, then you load it back

[grim locust]

thanks for the patience man, and sorry again for confusion.

[daring mulch]

Why does nobody care about my idea for deontological governance for agentic AI

[frank bough]

is there a software for mac where i can load in my index and pth files to inference?

i mean without python or terminal use

[proper goblet]

Hello

[polar flax]

@still nacelle deleted for possibly copyrighted content (music)

[still nacelle]

alr

[wanton furnace]

Is there any way I can use the voice model to voice chat live?

I want to try to speak my favorite character's voice when i voice chat with my friend

[echo inlet]

you can. like use voice model in chats by pairing it with a real-time voice changer. good luck!!

[full roost]

Hey im wondering what's the best AI API i can get for free at the moment?
I'm trying to build my first AI project, probably a chatbot, or a schedule maker.
I only own a personal laptop, nothing crazy, please have mercy on my lil buddy.

[frank bough]

how many epochs you guys recommend on applio with a 45 mins dataset?

[neon mesa]

hey guys

i wanna download ai voice

for the app called voice . ai

[lean viper]

@hidden grotto

[hidden grotto]

:wave: @lean viper, 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

[minor blade]

Don't use voice.ai, that's a paywalled program that also runs on the background of your PC.

-realtime

[pine acorn]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[minor blade]

Read the deiteris guide instead @neon mesa and for further help, go to the #✨│ai-help channel

[thorn mountain]

hi

[hybrid temple]

is vonovox better tahn this

[fiery pilot]

Is it possible to change the accent of an audio track and not just the voice?

It happens to me that I am changing the voice of an audio track with RVC and the accent does not change, I want to change the accent, for example: Colombian accent to Argentine accent

If anyone knows how to solve it, I'll give $ for their services

[placid token]

hey siri

[chilly lake]

Yes, you can use an index from a different model to add/remove an accent.

[fiery pilot]

I wanna to change everything

[open gazelle]

Does someone know if the hinamod inference for covers still working?

[earnest dune]

Help me pls

i cant pick devices

[daring locust]

-colab

[pine acorn]

📒 Google Colab Notebooks

Google Colab is a Cloud (Remote Good PC) Service. While the Free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

• **Applio**

by IA Hispano
Google Colab

• **RVC Mainline**

by Hina
Google Colab

• **UVR5 NO UI**

by Eddy
Google Colab

• **UVR5 UI**

by Eddy
Google Colab

• **Wokada Deiteris Fork**

by Deiteris & Hina
Google Colab

• **Hina's Modified Original Wokada**

Google Colab

• **RVC-AI-Cover-Maker-WebUI**

by Shiro & Eddy
Google Colab

• **FaceFusion UI**

by Nick088
Google Colab

• **FaceFusion NO UI**

by Nick088
Google Colab

• **Music Source Separation Training (Inference)**

by Jarredou & Makidanye
Google Colab

[uneven briar]

I need someone who is proficant in ai coding who can make a multimodal aI. i want to make a ai smart ar glasses and i need a team of pepole who can help. i will offer a incentive who ever helps me create the protototype gets 2% equity of the company. needs to be from united states i want someone specifically from California but if not its ok

[frank bough]

How much worth is your company yet ?

[stiff lark]

How do I change my voice for discord?

[vapid quarry]

I’ve always wanted to hear Arthur Morgan sing fool by Frankie cosmos

[sick thicket]

Whats the target LUFs for the RVC dataset?

I asked ChatGPT and it said -20

[covert lake]

heyy

[somber hornet]

🎉 | Nick088 by Weights leveled up!

[covert lake]

how yall doing?

3?

[covert lake]

if u need help theres #1192011222023950368

[night lake]

hi :3

[sick thicket]

do u know the target LUF for RVC dataset

[chilly lake]

experiment, experiment, experiment

[sick thicket]

spotify music is -14

the vocals within the track are often around -18 - -22

so imma try -18

I dont want the dataset to be low in volume atleast

it needs good dynamics

[chilly lake]

I used -16 in my test

[wide canopy]

yoooo

[sick thicket]

Imma use Fabfilter Pro L-2 and then level it up to -14 lufs and try to get the peaks up there

[sick thicket]

hows ur workflow for preparing the dataset

like idk if I should do Audacity and then do remove silence

[chilly lake]

audacity, trim silences to 0.25s, make sure no clicks or pops, then normalize to target lufs

[sick thicket]

ok

[chilly lake]

maybe a denoise before all that

that should get rid of more junk

[sick thicket]

the vocal is clean

i have 12 min of dataset

[tacit widget]

What are yall's thoughts on a model / workflow that would turn image(s) into pixel art

Any recommendations or ideas?

[broken raft]

-colab

[pine acorn]

📒 Google Colab Notebooks

Google Colab is a Cloud (Remote Good PC) Service. While the Free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

• **Applio**

by IA Hispano
Google Colab

• **RVC Mainline**

by Hina
Google Colab

• **UVR5 NO UI**

by Eddy
Google Colab

• **UVR5 UI**

by Eddy
Google Colab

• **Wokada Deiteris Fork**

by Deiteris & Hina
Google Colab

• **Hina's Modified Original Wokada**

Google Colab

• **RVC-AI-Cover-Maker-WebUI**

by Shiro & Eddy
Google Colab

• **FaceFusion UI**

by Nick088
Google Colab

• **FaceFusion NO UI**

by Nick088
Google Colab

• **Music Source Separation Training (Inference)**

by Jarredou & Makidanye
Google Colab

[chrome seal]

I want RVC of traning my voice to AI like Easygui

[solar torrent]

Good to know. There's Applio the RVC.

[chrome seal]

But i can't use Applio in Zero

[silk fog]

What happened to Weights GG?

Did someone take down Weights GG?

[solar torrent]

Weights.gg has been changed to Weights.com.

[silk fog]

Oooooh thank goodness 😮‍💨

[solar torrent]

Is that a problem? The website is still there, although the website can be down sometimes due to traffic issues.

[solar torrent]

Some Stable Diffusion checkpoint or LoRA models can make pixelated image from start. Downscaling and then upscaling into pixelated image would work.

[elder willow]

Just ask chatgpt or gemini to make an image to pixel art converter for you (in python or html5 or something).

[tacit widget]

..

Bruh

[elder willow]

It works o.o i did it

[chilly lake]

turning an imagine into a pixel or ascii art does not require AI

[empty pivot]

bruh I forgot which was ilaria's user, I wanted to dm her for reasons

nvm found it

[frank bough]

anyone got a working colab link for RVC2 (no applio) ?

[mint yacht]

i havent made a model in so long i forgot how to make one

[queen kernel]

Broh, you are a model maker

[polar flax]

if not sure and it sounds fine as is, the best bet is to leave as is and not mess up with dynamics by compressor/limiter, then try see the results

[atomic escarp]

yo i need help finding a program, does anyone know how people make those vids where they take an image of someone and they replace the person in the original video with them, and it mimics their movement n everything?

can send a example of what im talking about in dms

im tryna make some brainrot shit 😭

[echo inlet]

hmm this those deepfake-style videos where a face is swapped and animated to match the original person's expressions and movements?

[atomic escarp]

yes but not just the face like the entire body and everything

can dm an example like i said

[elder willow]

Fr

[jaunty pond]

guys, what's the number of "Epochs"?

(the bigger the better or worse?)

[minor blade]

None of both.

All will depend on the tensorboard and the voice itself because overtraining exists.

[jaunty pond]

i'm all ears than

[minor blade]

In fact, another way to find the best epoch off your model is using your ears and testing various pth and deciding for the one you like the most.

More info regarding training can be found on the guide: https://docs.aihub.gg/

Also, for further help use the #✨│ai-help channel.

Not this one.

[jaunty pond]

i found they range from 40 to 300 or even 750

[jaunty pond]

yeah yeah

so it's not a good measure and i have to test out specific examples myself

[minor blade]

Either testing with various checkpoints (pth's of your model) or observing the tensorboard.

[jaunty pond]

no idea what thos are

i will just pick whichever sounds better

[minor blade]

I will suggest you to read the guide then.

[jaunty pond]

is it the graph?

with latency and such?

[daring locust]

-huggingface

[pine acorn]

🤗 Huggingface Spaces

• **UVR5 UI**

Huggingface Space by Eddy and Ilaria

• **Ilaria RVC Zero**

Huggingface Space by thestingerx

• **RVC⚡ZERO**

Huggingface Space by r3gm

• **Applio**

Huggingface Space by IA Hispano

• FaceFusion

HuggingFace Space by Nick088

[dreamy hare]

check

[finite meteor]

what's a decent processor to run realtime vc??

[white parcel]

any modern cpu could do i think

[obtuse quest]

I need a voice model

[supple stream]

#🔍│find-models

[elder willow]

what are some free good Ai image generators that can generate high quality detailed pictures?

for example if i want multiple angles of a duffle bag and for it to be detailed and high res.

[open gazelle]

Can someone explain me how does this work?

[chilly lake]

do you have multiple cuda devices in your pc?

[open gazelle]

Nope!

[chilly lake]

then it is 0

[queen kernel]

Which gpu do you have

[elder willow]

is there no download stuff?

[queen kernel]

I mean I have to know about your PC specs so that I can know that your PC is capable or running that or not

[elder willow]

amd radeon RX 5700 XT

didnt know hardware matters for no download stuff

[queen kernel]

Like what ?

[elder willow]

asking for Ai that i dont have to download and run on my pc

something more simple like a website?

[queen kernel]

Ahhh, my bad

There are many websites but they all require credits

They have limitations

[elder willow]

i mean if they give high quality results i dont mind spending 10bukcs

[queen kernel]

What do you think about mid journey

[elder willow]

not familiar with it

you reckon its good?

[queen kernel]

You have never listened about mid journey?

[elder willow]

probably generated a cat image in the discord a year ago 🤷‍♂️

that's it

[queen kernel]

https://www.midjourney.com/explore?tab=top_month

Go with it if you like it

[elder willow]

that's good thanks

know a good free one i could compare with?

[queen kernel]

Check samples on Civit.ai

It's an open source platform but requires downloading and stuff.

[elder willow]

👍

[neon swallow]

Hello! I'm looking for an api that i can send images to and a prompt, and will generate me an images based on those. anyone knows a tool providing that?

[fresh rain]

i just made a song w a mr beast voice changer

LOOOOL

[covert lake]

helloo

[covert lake]

nice

[night lake]

hello

[covert lake]

deltarune

today

[night lake]

cool

what is that

[covert lake]

https://tenor.com/view/deltarune-today-today-deltarune-realese-dance-gif-480092671422614917

[covert lake]

the videogame

the canon parallel universe of undertale

[night lake]

ah

[covert lake]

they just released chapter 3 and 4

chapter 2 was released in 2021

[night lake]

coolio

how many chapters are there total?

[covert lake]

7..

[night lake]

bruh

[covert lake]

yes my grandkids are gonna play it

[night lake]

yeah that game is gonna finish in 2050

[covert lake]

chapter 1 was in 2018

[night lake]

oh

yeah my kids kids are going to be playing that

[covert lake]

but now they got a better indie squad

because before it was only 1 dude

[night lake]

now its what?

3 dudes

[covert lake]

i think 6 tbh idk 😭

its an indie bro

[night lake]

wild

[frank bough]

Anyone got a solution for this. Trying to convert with RVC on Mac M1 Traceback (most recent call last):
File "/Users/jlapping/.pyenv/versions/3.10.11/lib/python3.10/site-packages/gradio/routes.py", line 488, in run_predict
output = await app.get_blocks().process_api(
File "/Users/jlapping/.pyenv/versions/3.10.11/lib/python3.10/site-packages/gradio/blocks.py", line 1434, in process_api
data = self.postprocess_data(fn_index, result["prediction"], state)
File "/Users/jlapping/.pyenv/versions/3.10.11/lib/python3.10/site-packages/gradio/blocks.py", line 1335, in postprocess_data
prediction_value = block.postprocess(prediction_value)
File "/Users/jlapping/.pyenv/versions/3.10.11/lib/python3.10/site-packages/gradio/components/audio.py", line 349, in postprocess
file_path = self.audio_to_temp_file(
File "/Users/jlapping/.pyenv/versions/3.10.11/lib/python3.10/site-packages/gradio/components/base.py", line 325, in audio_to_temp_file
temp_dir = Path(self.DEFAULT_TEMP_DIR) / self.hash_bytes(data.tobytes())
AttributeError: 'NoneType' object has no attribute 'tobytes'
2025-06-04 15:10:38 | INFO | httpx | HTTP Request: POST http://localhost:7865/api/predict "HTTP/1.1 500 Internal Server Error"
2025-06-04 15:10:38 | INFO | httpx | HTTP Request: POST http://localhost:7865/reset "HTTP/1.1 200 OK"

[covert lake]

#1192011222023950368 and elaborate

[chilly lake]

you already asked 5 hours ago. Do you even read responses?

[molten seal]

whats the best model for video generation man

[chilly lake]

Wan with some tricks or FramePack (Hunyuan basically, with unlimited generation size)

Use Comfy for WAN or SD.Next for FramePack. The standalone FramePack has issues and dont have proper offloading and quantization, but in SD it works great.

[chilly lake]

that's for free stuff, or there's Gemini Veo3 that is mind-blowing

[molten seal]

bet bet

[wet sage]

guys? how can I upload fav voice?

[frank bough]

I do read responses but it didn’t help 😂

[chilly lake]

was thre a different error?

[earnest gorge]

Hey ai

Can I be the owner of this discord server

[night lake]

yes

[wet sage]

@mint yacht

@unique pawn is it possible for you to make mecha aatrox model?

[unique pawn]

Why me specifically

[wet sage]

cuz you have good voice overs

esp. ultron

like sounds very good

guys?

how can I use

w okada

in voicechat?

[polar flax]

https://tenor.com/view/arona-arona-blue-archive-blue-archive-dance-cute-gif-5770112494656837935

[polar flax]

the "chapters" feel like deltarune sequels

[sterile marsh]

هاي

Hello

[drowsy plinth]

This voice model can only be used in voice.ai or there’s other programs you can use

I’m asking this because voice.ai on discord keeps cutting off everything I run a game once I exit out of the game discord voice.ai becomes good

[polar flax]

I don't recall there are voice models locked to it (or perhaps you mean kits.ai?)
nevertheless both voice.ai and kits.ai are utter garbage and paywalled with ridiculous pricing

[still sigil]

@clever fern no advertising and please keep the chat in english

[drowsy plinth]

Oh ok, I was trolling friends with a male voice on discord but once I decided to play a game and have the voice on it will cut off so I was trying to look for another voice changer that I can play on the background and play games at the same time

Also I was using a voice model I found so I wanted to know if I can use it with another changer program

[solar torrent]

The only good voice changer program is W-Okada, not Voice.ai.

[drowsy plinth]

Can I use the voice models from here in there?

[polar flax]

please take a look on the voice changer guide, depending on your gpu spec https://docs.aihub.gg/rvc-voice-changer/local/deiteris-w-okada-fork/

[drowsy plinth]

can i use the same voice model from here?

[polar flax]

as long it is downloadable

the ones in https://discord.com/channels/1159260121998827560/1175430844685484042 and weights.com should be so

[thick tapir]

Nya

[drowsy plinth]

thank u for the help

[echo inlet]

hii, are u pertaining to using an image as a reference style or pose, then dropping a prompt for the ai to generate? i just use savro for that

[north pewter]

is there a chance we're getting RVC3?

[polar flax]

not before GTA 6 release ffs, and not even the witcher 4 or cyberpunk 2

https://www.youtube.com/watch?v=aorRfK478RE

[north pewter]

amazign

[solar torrent]

I have a job. Not really necessary for me to do that.

[grim locust]

yoh, just want to thank Noobies for the code. My Zonos is alot faster now that i dont have to regenerate the speaker embedding everytime i generate. thanks alot man

[high sun]

.

[chilly lake]

sus

Deepfaking a video of a minor saying things they did not say has a big potential of seeing the sky thru a metal bars for a few years.

Then she has 40 minutes to record the video herself.

[still sigil]

i mean they didnt say that the person in the video is their sister

did the person in the image consent to having themselves deepfaked and lipsynced?

no?

proof

deepfakes are already mainly used illegally

you wont find any help here

great

find it elsewhere

[chilly lake]

It is the internet, there's no proof for anything you just said.

for one legit deepfake there's probably 999 others with people doing nasty things

[wary vale]

-realtime

[pine acorn]

💻 Local Realtime RVC

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

• Wokada Deiteris Fork

Most suggested. GUIDE

• Original Wokada

ONLY the latest alpha comes close to the Deiteris Fork performance, older versions in youtube tuts are way worse. GUIDE

• **RVC GUI Mainline Realtime**

Unavailable, the guide is outdated and the program is worse compared to the ones above, and much less updated

[covert lake]

helloo

[dusty dove]

Gay

[echo inlet]

uh u looking for like making an ai model with high quality image, then start manipulating from there?

[frank bough]

anyone can train me a 25 mins dataset to 300ep on RVC2 ?

or applio if it's fast on your local

[night lake]

Rvc is also a deepfake

😎

[rich sable]

hey everyone - i'm on mac can anyone pls teach me how to download a local unlimited free ai video generator (idk like comfy ui) im new to this so if you could pls help

[chilly lake]

I think it's been way past that 40 minutes limit dude 🙂

[rich sable]

lol bro i figured out a solution

and this is a whole different requesr

*request

[desert kayak]

came across this! https://www.linkedin.com/posts/shaneak_you-will-be-an-ai-engineer-in-the-next-5-activity-7336408721759117312-6MtC seems like a step above bolt/lovable

[chilly lake]

another advert to a dumb vibe coding site

truly.. 2026 will be a year of vibe coding

but I think 2027+ will be a year of actual well paid engineers fixing all that garbage

[olive otter]

anybody here has experience with using ai for reverse engineering?

[stray drum]

Has anyone here ever used RVC to map style onto TTS?

[chilly lake]

confused unga-bunga

[stray drum]

style being emotion

@chilly lake

happy to explain further over a call!

[chilly lake]

if you need tts that does produce emotional reads, you need to get that tts.

with or without zero-shot voice cloning

if it does not provide zero-shot cloning, you can use RVC to replace the voice

but RVC has nothing to do with emotions or style transfers

[stray drum]

Take the OBS plugin for RVC for example
Imagine Obama's voice is a model option
When I speak into my mic, the audio out is Obama's voice
When changing to Obama's voice, if I scream, will I scream like Obama?

[chilly lake]

Not likely, most models are not trained with such wide range

[ripe iron]

it dont even work coz my gpu ios ass

hd graphics 4600

imma need a new pc

[olive otter]

have you worked with sesame?

[stray drum]

i have not @olive otter

are you available for a quick chat? I'm assuming you're talking about csm-1b

[elder willow]

Hi, could I get some help please? When I run 'pip install -r requirements.txt', I get this:

Collecting numpy==1.23.5 (from -r requirements.txt (line 8))
  Using cached numpy-1.23.5.tar.gz (10.7 MB)
  Installing build dependencies ... done
  Getting requirements to build wheel ... error
  error: subprocess-exited-with-error

  × Getting requirements to build wheel did not run successfully.
  │ exit code: 1
  ╰─> [32 lines of output]
      Traceback (most recent call last):
        File "C:\Users\titoi\AppData\Roaming\Python\Python313\site-packages\pip\_vendor\pyproject_hooks\_in_process\_in_process.py", line 389, in <module>
          main()
....

[magic carbon]

anyone got an mafioso or bluudude voice?

model

[olive otter]

yes

[chilly lake]

dont use 3.13

[obtuse plaza]

Good night

What app/site do you guys recommend for RVC on mobile?

I used colab sites a few years ago, but for some reason they don't work anymore

[tawny oracle]

are there any good bots for making vst plugins, was just wondering

[junior thunder]

is there any good voice changer that i can use the models with?

[still sigil]

https://tenor.com/view/lesbian-kissing-lesbians-girls-kissing-kiss-girl-smooching-lesbians-gif-7857077274454131314

[junior thunder]

WHATT

[still sigil]

im a straight male and lesbians are hot

[junior thunder]

i agree

[echo inlet]

hmm there are a few that gets the job done, but it is better to be used for testing stuff out only

[dapper drum]

Any tips with izotope rx 11

[coral vigil]

be a gay

[polar flax]

yagami

[elder willow]

my voice changer client demo is always freezing

[frank bough]

question: could it be that applio is more precise when it comes to singing powerful and soft? i feel like my model i trained with applio can belt but also sing soft. My RVC models tend to do something in between.

[lyric anchor]

Hey all - just joined here so apologies in advance for any repetitive questions.
I'm pretty new to AI, so is there any material / videos that anyone could recommend,
I'm specifically struggling with getting ChatGPT to recall and give me an accurate time.
Anyone had similar issues or any advice on how to resolve?
Thanks in advance

[chilly lake]

why are you using a demo from 3 years ago?... probably mixing the device types

[elder willow]

It says demo at the top

[chilly lake]

it is a demo

from almost 3 years ago

[elder willow]

Which is the most updated version for windows 11

Could you kindly send me the link

[chilly lake]

https://rentry.co/forkvoicechangerguide

or Vonovox

https://github.com/dr87/Vonovox

[elder willow]

Iam using huggingface

[elder willow]

can u come in ur dms rq

[chilly lake]

okay, that's little beter, I guess

try running i in server mode instead

[elder willow]

so should i use the one iam using rn or the vonovox

[chilly lake]

with 4060 you can try vonovox, it is the most up to date voice changer

and fix your chunk and extra values

[elder willow]

wym "fix"

[chilly lake]

https://rentry.co/forkvoicechangerguide#finding-my-own-settings-for-chunk

[elder willow]

ther eisnt anything green

[chilly lake]

In your beta, yeah

[elder willow]

alr imma download vonovox now

thanks for guiding me man

my englis really isnt the best

[elder willow]

another issue look in ur dms please

[chilly lake]

it takes ~10s to start

[elder willow]

can i make my output somehow louder

[chilly lake]

not my project, have not really used it

[quiet leaf]

help, I installed the voice changer and virtual microphone correctly in the voice changer, I also installed the virtual cable on the microphone correctly in the discord, I start, I say, it doesn't work, but when I turn on the loud video, the voice changer perceives it as a voice, and changes the voice to the video, in general, instead of the microphone sound, the voice changer changes the sound of headphones

and now, it captures both the sound and the microphone, what should I do?

[barren cradle]

I am looking person to rvc my voice to the song..i make album so i need..pm me if you intersted

[lyric anchor]

Hey all - just joined here so apologies in advance for any repetitive questions.
I'm pretty new to AI, so is there any material / videos that anyone could recommend,
I'm specifically struggling with getting ChatGPT to recall and give me an accurate time.
Anyone had similar issues or any advice on how to resolve?
I'm also interested to understand if / how I can link up my various platforms to create an autonomous set of AI agents who need minimal human supervision or direction?
Thanks in advance

[river verge]

DIlly ding, dilly dong! A new RegalHyperus drum model just released!
Giant Steps (Drum model no. 608)

[covert lake]

elaborate in #1192011222023950368

[covert lake]

#1192011222023950368 for asking how to do it yourself

[covert lake]

share ur pc gpu, the tut link and what u want to do in #1192011222023950368

[loud jackal]

.

[rose cobalt]

Guys I just made a Lil Mosey 2019 Best era model now if u want to check it out

[humble sedge]

what is best momy voice

[mighty ermine]

Hello how can ı make my own voice model?

[left lotus]

Can someone tell me how to get the Voice changer pls!

[shell crag]

same here!!!!!!!

[lost narwhal]

Hi. Could someone please make a Voice Model of David Kawena (Lilo and Stitch)?

[boreal tinsel]

helloo, can someone tell me how can i download w-okada voice or is there a better option??

🙏

[dusty atlas]

man the weights website is down

[bold rampart]

This is cool... The first audio is the original audio, and the second is a piano version made by Suno AI

[echo inlet]

hi, from what i know, u can find it usually on GitHub? but it is said to be quite difficult to manage. for other options, well there are beginner-friendly ones like XTTS, Bark, SavroAI, and RVC for more control that you want

[thin granite]

for the rvc how do i stop the robotic parts the mmmodel has sometimes when im speaking

[gloomy copper]

heyy

what mic is best for voicechanger?

[rustic steppe]

Hi, could someone help me to edit a photo?

[mystic roost]

Event Mechanics

:tada: How the Raffle Event Works?

1. Create images using the Voyages app and copy the share link.
2. Post the link in the raffle entry channel (ask staff if unsure).
3. Each valid link = 1 ticket. Multiple valid links give more tickets.
4. You can check your entries with /tickets, and view rankings with /leaderboard.
5. Duplicate links, fake links, or misused links may get you banned from the raffle.

:gem: Commands you can use ~

• /tickets – View your (or others') ticket count and rank
• /leaderboard – See the top participants in the current raffle

:trophy: Winners are picked randomly, but your chances increase with more tickets!

-# Please leave an awesome review if you love our app on Play/App Store

[young grove]

🤖

[elder willow]

can someone help me

[upper bobcat]

yo how do i use an ai voice

like how do i change a voice with ai

[minor blade]

Use the #✨│ai-help channel.

[spare walrus]

How do i make a ai cover on my phone

Without weights bc it sucks

[humble lily]

hi how can i download the model in www.weights.com
www.weights.com sitesinden modeli nasıl yüklerim

[covert lake]

#1192011222023950368

[covert lake]

we dont allow requests

You can search rvc ai voice models at:

• #1175430844685484042
• In #🔍│find-models , Do /find with @hidden grotto
• https://weights.com/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice 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:

• make it yourself with our docs guides

[hidden grotto]

:wave: @covert lake, 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

[covert lake]

**elaborate in #1192011222023950368 **

[covert lake]

works fine on my end

[covert lake]

its better to use the docs

[covert lake]

elaborate in #1192011222023950368

[covert lake]

**elaborate in #1192011222023950368 **

[covert lake]

**elaborate in #1192011222023950368 **

[covert lake]

use #1192011222023950368

this server isnt made for looking for jobs @west island

[covert lake]

speak only english and use #1192011222023950368

[night lake]

Why hello there Mr Roy Sebastian

[shell verge]

how can i use this on vr?

[covert lake]

**elaborate what you want to do in #1192011222023950368 **

[shell verge]

there

[bright siren]

@covert lake apologies to intrude just wanted to ask were u born a goat? or did overtime u just become ga ga ga goooaaated?

[covert lake]

thanks lmao

@golden moss please dont promo here

[exotic cliff]

any1 knows where i can find other voice models that are not listed on this dc

[dusty atlas]

nevermind

[chrome seal]

How i use Applio RVC to train my voice

[rough sky]

whats the best kanye voice model in your guys opinion

[queen kernel]

Ask in #✨│ai-help

[spice garden]

what is voyage contest

[echo sage]

can anyone help I use okada voice changer but my voice lags a bit and mumbles a lot

[still sigil]

you create 3 images a day on voyages for a chance to win the prizes

[covert lake]

You can search rvc ai voice models at:

• #1175430844685484042
• In #🔍│find-models , Do /find with @hidden grotto
• https://weights.com/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice 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:

• make it yourself with our docs guides

[hidden grotto]

:wave: @covert lake, 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

[covert lake]

#1192011222023950368

[covert lake]

elaborate in #1192011222023950368

[exotic cliff]

thanks

[fallen heath]

Hi! I'm looking to build a headshot app. What image generation APIs do you recommend? I know chatgpt 4-o's API for image output isn't live yet so curious what people are using - especially other headshot apps e.g. https://www.headshotpro.com/ and https://www.aragon.ai/

[solar torrent]

A goat or GOAT? Neither of these. I'm a normal man.

[covert lake]

They weren't talking to you tho

[alpine bone]

i

hi

[covert lake]

Hello

[alpine bone]

where re u from

[covert lake]

italy wbu

[alpine bone]

usa

[honest silo]

Hello

[solar torrent]

Hi.

[honest silo]

What is done here?

[solar torrent]

This chat is about AI related topics. The normal conversation goes to #✦│chat.

I'm not down. I'm listening to music.

[radiant hamlet]

would weights accept it to train if i add a 1 hour video

[crimson ember]

hello can you help me with RVC plsss

I watched some vids on ytb and they said to download the voice-model and then extract it for the model

then putting the model in the model file of RVC and it should be done

but when I when I downloaded it it was just a mp3 file and I can not extract them for model file

so what should I do

please gimme some advices

[queen kernel]

#✨│ai-help is the right place to ask

[graceful cypress]

hello kan somoene help me wicht making a game?

[broken vapor]

where can i get RVC?

[supple atlas]

github

[covert lake]

#1192011222023950368

@midnight shadow Don't request models, its not allowed

[lost narwhal]

But you can make requests on Reddit

[broken vapor]

thanks

[sudden karma]

Can someone help me set up the rvc?

[exotic cliff]

whats the problem

[sudden karma]

it doesnt work

like the voice doesnt change

[queen kernel]

#✨│ai-help is right place to discuss your issues it RVC

[novel glen]

I have a system with the Ryzen AI MAX+ 395, been trying to get an AI model setup utilizing Whisper DML to transcribe voice to text from lectures and then organize it into summarized notes.
Problem I've run into is that whisper.cpp DML does not fully expose the NPU, ONNX runtime doesn't target the latest NPU yet either, and AMD's IpuMcdmDriver stack is not accessible with public APIs

Anyone know anything else that might work for that task?

[echo inlet]

Hi, i'll say that NPU support do still feel half-baked in a lot of AI tooling right now. so if Whisper DML isn’t fully hitting the NPU, maybe try running Whisper through OpenVINO or look into Vulkan-powered backends? they serve good as workarounds. results still depends though since the frameworks of these aren’t always finely tuned for Whisper’s architecture

[novel glen]

Alright I'll check it out

[bold rampart]

This voice is AI generated

Prompt:

Okay [laughs] so I asked my friend to tell me the most unfunny joke he knows, like I wanted something so dead inside it circles back to being hilarious, right? [swallows] And he just goes, totally serious "Why did the chicken cross the road?"— [laughing] I'm already bracing myself—and then he hit me with it [snorts] "To get to the other side." [starts laughing] Brooo—BROOO I don't know WHY that sent me!! [laughs harder] Like the chicken had NO PLOT, no twist, just raw determination!! [laughs hard]

Another one, this one is more realistic, but uses an official voice:

If you want, this is one of the generations I considered failed:

[naive python]

/image

/image маленький бабиджон

sorry guys idk how this working

[still sigil]

do it in #🤖│bots

[open bolt]

hey yall
do any of you use AI in your projects?
i want to integrate AI into my project
but i dont know how to do it for free

[novel glen]

Does it run on the NPU or the GPU

What I require is low response time, low latency

[gloomy bay]

Hi

I have a stupid question

For ai workload 8gb 3070ti or 16gb 3060ti ?

Or should I go with amd

What abt amd ?

Do they have any release date for it ?

Thanks for the help

[polar flax]

aren't both 8 GB unless you mean the latter 4060 Ti?

[gloomy bay]

My bad

[novel glen]

In any case, how does it transcribe live audio from a microphone to text and then organize it into notes? Is that something it can handle?

may isn't good enough

[shy smelt]

why i cant change index ?

everytime i use a model its keep cutting out and it doesn't get all words right when i talk !!! even though the recordings made here before i download them sound so clear and nice

[covert lake]

#1192011222023950368

[covert lake]

please elaborate in #1192011222023950368

[hexed hamlet]

nick help mee

[covert lake]

please use #1192011222023950368

[hexed hamlet]

yeah

sorry I just reply pinged you twice

[solar torrent]

No excuse. Use a proper text channel for help, like #✨│ai-help or #1192011222023950368.

[hexed hamlet]

The previous community text channel is removed for no reason

[stiff hare]

hi guys

[hexed hamlet]

I made the ticket and said this

[queen dragon]

Has anybody done a model with a voice that's vocoded?

[barren mauve]

no?

but i wich

[terse maple]

hello who can help me im new to ai need some advice

[inner pasture]

where do i put ai songs?

[dusty atlas]

man Generate Image is still taking too long. no ETA and 3 hours passed and the image is still not ready it's not fair

[oak anchor]

Hey does anyone know a really realistic image generator?

I heard about midjourney but i want to be sure it will be really reaslitic before i spend money

[oak anchor]

What do you think the next best thing is for a noob like me, i used chatgpt to generate a image but it was super ai looking

I have a gaming pc with 16gb ram

3050 graphics card

16gb ram

[still sigil]

vram

[oak anchor]

Idk what that is tbh let me check

I have 8vram

sd1.5?

Does it help if i say i need the ai for commercial purposes, like lets say i have a picture of a person and i want them to hold a product im selling

is there an ai specializing in that?

Ah i see.

i see

So the guy im competing with either hired somebody to do it or is really good at it

He took a picture of a ramdom dude on yt and made him promote his product all ai

and the video went viral

What do you mean?

what if i buy midjourney

I heard its the best

That looks hella complicated

[dusty atlas]

it's not fair! the image is still not ready and it's been 4 hours it's still taking too long and no will respond to my feedback and i just want my image to be generated

[bold rampart]

BRUH... ElevenLabs is too funny...

[queen dragon]

It can generate decent screaming? Wow maybe it's getting better than I thought

[bold rampart]

Generated by Suno AI

Yeah, and also, I paid for Suno, so I can use my music commercially

It's $10

Me just trying to create Trap Nation-like songs

Don't worry, it's not copyrighted

EDIT: $10 per month

YEAH! THIS IS GOOD!!!

Even better...

[bold rampart]

With vocals

[eager path]

the apples claims on the ai not being able to think is clearly absurd. As in humans not communicate and act as the AI does, get the data, process it, and give an output. Literally the same as most of the people, for they don't really think that much from indulding in hedonistic activities.

[still sigil]

didnt know apples could think

dont forget what happened the last time an apple gained consciousness

[potent cradle]

what are some of the most convincing/best sounding voice models out right now?

[hexed hamlet]

convincing

[boreal cargo]

Anyone can help in support vs, trying to run a py file

For some reason python (name.py) dosent work with it on desktop

all dependices were installed.

[crude star]

hello guys, im in need of some assistance T_T
ive recently been asked to create an animated short based on a script given to me but idk how to go about it since ive only ever used chat gpt for essay generation and ive tried my hardest to look for free models that can help me with this but am unable to find one that works
does anyone have any advice or tips that can help me with my current scenario

[crude star]

totally about 15-20 seconds maybe 30 if possible ( by putting together 5 second shots each)

ive managed to generate a storyboard of the shots im happy with but the image to video generation is the part ive been stuck for 2 weeks on 😭

yeah ive seen incredible output from veo 3 but the thing is im a student 💀 so i genuinely cannot afford to pay for models cuz i get nothing in return from submitting this video output too T_T, ik i sound insane but im just telling u my situation 😭 sorry if this comes out the wrong way

[solar torrent]

You can use alternatives to that one instead, some are cheaper, although some features are limited for free accounts.

[crude star]

that sounds interesting, are there any ones you suggest, since veo 3 actually isnt available where im from atm

oh those arent required, i just want the characters in my image to move a bit lol

unfortunately i only have 6 gb vram with 16 gb ram 😭

ive tried comfy ui before but i couldnt get anything usable

fair, thats understandable lol

[solar torrent]

What? Voice dubbing TTS and video generation don't have to be in one at all. You can do TTS from another website, but you'll have to manually edit them into one video.

[crude star]

i really appreciate the advice though :D, ill just have to keep looking for online models or give up lmao

tqsm! :D

[solar torrent]

Crazy. But I find cheaper alternatives to be funny.

[pulsar schooner]

which ai girl voice model is the best?

[tardy gorge]

yo, does anyone know any good ai image editor thats free?

[covert lake]

please use #1192011222023950368 and tell ur pc gpu

[tribal stirrup]

Sorry for being in a unrelated channel, but is there any way to request voice models in case someone's curious enough to make one? It's been a while since I visited Ai Hub and I could not find a decent voice model of a specific character in the original language, I could not find it on Weights.gg (which is where I posted/made my more recent models) nor voice-models.com or via the voice model search.

[hard pendant]

@stark escarp pls unmute me

Unmute me

@stark escarp

I sorry

@stark escarp

Pls forgive @stark escarp

PLS

AM SORRY

PLS @stark escarp pls i am sorry

Ok

@stark escarp am sorry i wont do that again

JUST PLEASE UNMUTE ME

I WONT DO IT AGAIN

I PROMISE 😢

@stark escarp where u streaming

@stark escarp what your channel

I will sub

[river adder]

@hard pendant stop spamming or we'll have to mute you for a while. You are annoying other server members

[queen kernel]

@stark scarab sorry to bother you, but I have some suggestions for UVR 5 UI. I'm using it and it works perfectly fine but I want that when I do batch separation, it put instruments and vocals in different folders. Reverb and non reveb vocals 8n different folders and so on. Is it possible?

[stark scarab]

Oh, u want like a filter like... let's suppose u separate 10 songs, so u want all instrumentals in one folder and all vocals into another folder?

[queen kernel]

Yes you are right. I think it will be very helpful. Because after separating vocals, I have to de-reverb them. After that de-echo and de-noise so it's very frustrating every time moving every file into different folders

[stark scarab]

Added to my TO-DO

[queen kernel]

Also is it possible to just check ✅️ whatever I want. Like for preparing dataset I don't need instruments or reverb or noise files.

I remember in uvr 5 gui there was check boxes and we can select instruments or vocals only for output

[stark scarab]

that's already added

but not like UVR5 does

U need to type the stem u want

check this:

https://github.com/Eddycrack864/UVR5-UI/issues/52

stems names are placed on the filename btw

u probably knows

[queen kernel]

I thought having check boxes will be more convenient

[stark scarab]

It will be harder for me but I'll try

[winter flicker]

What to do if neronka does not work rtf 0 and , vol: 0.0000

[stiff lava]

hello

i need some help

im looking to create an ai influencer ( im trying and im overwhelmed ) with comfyui, can someone help me with a workflow that could run on my laptop? low vram ( 6gb vram ) rtx 3060 laptop

text to image workflow + lora stacker

i have no idea where i can find a good workflow

i ve been looking for at least 2 hours

on civit ai

and i find corrupted workflows

they arent working

and im feeling like i want to give up 😭

yea yea i know

but i dont have the workflow

i found this one

https://openart.ai/workflows/deckdiavalo/chroma-text-to-image-low-vram/f2gn3k0ncNxwCz4vAdz4

but idk

i will download to see if it works

i hope i dont waste my time

he said in description it is

he gave a checkpoint

idk

oh

god damn

so what would u recommend me?

let me search one on civit ai

you have a point

i ll try experimenting

i tried to remake it

where can i share a photo?

im not allowed there

i ll go to eat

meanwhile

what s a decent checkpoint to generate sfw/nsfw people photos with ai and lora too?

[queen kernel]

what this thing means ??

[queen kernel]

the same on tensor board ?

i see. thank you a lot

[queen kernel]

Okay.

Got it

[stiff lava]

where can i get it?

nvm i guess i found it

analog madness v40 from ckpt

???

[bleak radish]

yo anyone know how I can make ai music on mobile

[stiff lava]

suno ai

the best

[bleak radish]

does it let me use ai modules

I need one that will let me use the models.

[minor blade]

Nope, you can't use RVC models in Suno

[stiff lava]

oh

[crimson dragon]

yo, is there a guide for training ai for generating images? Preferably ai should be local

[cerulean cradle]

@pale lotus you are gay?

[pure ice]

Guys look at this https://youtube.com/shorts/ellbpuCgm_g

[glossy parcel]

are tgere any ai german women changers?

[austere dove]

can anybody help me

with the voice changer

it delay

[bold rampart]

Bruh, I currently have 5,870 credits in Suno, and each generation costs 10 credits, so I can create approximately 1,174 songs left

[void snow]

I need some help with a project ive been working on. i need some messy non structured chat logs or brain dumps notes anything.... and see if my ai can structure them into something usable.... the longer the better.

[echo inlet]

answer to this is reddit hahaha. both you're looking for (non-structured and brain dump notes) are there :))

[bold rampart]

Yeah, next month they will all disappear and go back to 0

[bold rampart]

Bruh, Google’s Gemini image generator is not NFSW restricted at all! I tried a few, of course I wouldn’t share it here, and yeah, turned out quite weird

[marsh stone]

Its real?

[bold rampart]

Yay! I now officially became a Weights Premium member

[mental bough]

what

[queen kernel]

Noobies explained

[mental bough]

yeah idk

[bleak radish]

anyone got a tut for mobile? with the rvc stuff

[queen kernel]

Sorry but external links or promotions not allowed in this server

[queen kernel]

For mobile, you have to use cloud platforms, such as kaggle or colab.

[bronze olive]

where can I find the ai image generator from this ai image generator channel: https://www.youtube.com/@inspiringdesignsnet/shorts

[subtle osprey]

what setting refers to making voice more understandable?

[lost kindle]

need a bit help here

for some reason i cna't paste images

https://colab.research.google.com/github/w-okada/voice-changer/blob/master/Realtime_Voice_Changer_on_Colab.ipynb#scrollTo=86wTFmqsNMnD the default way is giving some errors

[still sigil]

https://tenor.com/view/valsfavoritegifs-gif-16718928798951018101

[lost kindle]

Installing pre-dependencies...
ERROR: Could not find a version that satisfies the requirement faiss-gpu (from versions: none)
ERROR: No matching distribution found for faiss-gpu
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 261.0/261.0 kB 7.1 MB/s eta 0:00:00
Preparing metadata (pyproject.toml) ... done
Building wheel for pyworld (pyproject.toml) ... done
Installing dependencies from requirements.txt...
ERROR: Ignored the following versions that require a different python version: 1.21.2 Requires-Python >=3.7,<3.11; 1.21.3 Requires-Python >=3.7,<3.11; 1.21.4 Requires-Python >=3.7,<3.11; 1.21.5 Requires-Python >=3.7,<3.11; 1.21.6 Requires-Python >=3.7,<3.11
ERROR: Could not find a version that satisfies the requirement onnxruntime-gpu==1.13.1 (from versions: 1.15.0, 1.15.1, 1.16.0, 1.16.1, 1.16.2, 1.16.3, 1.17.0, 1.17.1, 1.18.0, 1.18.1, 1.19.0, 1.19.2, 1.20.0, 1.20.1, 1.20.2, 1.21.0, 1.21.1, 1.22.0)
ERROR: No matching distribution found for onnxruntime-gpu==1.13.1
Successfully installed all packages!

is there any way to fix it?

[still sigil]

ask in #✨│ai-help

[lost kindle]

thnaks

[mental bough]

-colab

[pine acorn]

📒 Google Colab Notebooks

Google Colab is a Cloud (Remote Good PC) Service. While the Free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

• **Applio**

by IA Hispano
Google Colab

• **RVC Mainline**

by Hina
Google Colab

• **UVR5 NO UI**

by Eddy
Google Colab

• **UVR5 UI**

by Eddy
Google Colab

• **Wokada Deiteris Fork**

by Deiteris & Hina
Google Colab

• **Hina's Modified Original Wokada**

Google Colab

• **RVC-AI-Cover-Maker-WebUI**

by Shiro & Eddy
Google Colab

• **FaceFusion UI**

by Nick088
Google Colab

• **FaceFusion NO UI**

by Nick088
Google Colab

• **Music Source Separation Training (Inference)**

by Jarredou & Makidanye
Google Colab

[rancid ravine]

Did this youtuber use Suno AI to make their content?

https://www.youtube.com/watch?v=bne99RE4DGg&list=PLd7kDGJ508yMBR8MKJz88B5-4gdW7-jfj&index=5&ab_channel=LEWKY

[normal fern]

guys,i have [Voice Changer] Pipeline is not initialized.

how do i fix dat

[bold rampart]

B.R.U.H: Google Gemini can generate copyrighted characters without refusing too

[normal fern]

rocky

do u know how to fix pipeline is not inilialized

[bold rampart]

Sorry no.

[queen kernel]

Ask in #✨│ai-help

[gilded moat]

what is the best female ai model?

[night lake]

this one: https://discord.com/channels/1159260121998827560/1341216399372062823

[nocturne sundial]

Is there a channel for AI music too?

Cause I tried making a song and the result went absolutely nightmare material

[calm crane]

Hey, I was wondering if running an LLM locally will reduce the effect it has on the environment? I hear a lot about how chatgpt has adverse effects on the environment and i don't want to be contributing to that

[nocturne sundial]

I am also worried about this, and as far as I know, DeepSeek has a lower footprint. Nonetheless, I am not really sure if running an AI locally would help as I suppose that cloud computers have optimised hardware for that...

[rare scarab]

anyone have a good french e girl voice

or just a good models

[dense terrace]

are there any voice changers that work fine with a terrible gpu

[daring herald]

Hello, I want to make a video of Lionel Messi greeting my father on his birthday. I already have the audio and video that I would like to use. Does anyone know of any free tools that I could use to make the deep fake?

[marsh canopy]

W move
comfyui but you need good gpu, if your gpu trash rent one for like 0.8$/hr on runpod [dot] io

[queen kernel]

If you are running an LLM model locally, you require a very huge amount of VRAM and You have to use multiple GPUs and they will produce a lot of heat so you have to cool down them before they get cooked. So you need to cool it on large scale for that you need active water loop cooling system. Also GPUs require a huge amount of power, to genrate power, we burn fossil fuels. So there is no way to save the environment. If you are really worried about environment just stop using Ai And stuff.
(Running small LLM models will give you wrong answers. So there is not any benefit to run small LLM models locally. If you want accuracy you must need large models for large models you need high computational power)

[calm crane]

aw man

[strong flare]

anyone know where i can find the text to speech used in this video https://www.youtube.com/watch?v=nmg0EYpxH-M its really common, just not sure where to find it

[queen kernel]

I explained everything in detail

[calm crane]

it's probably worth noting that i can generate images locally though

[queen kernel]

To genrate images you don't need LLMs

[calm crane]

i know

[velvet lichen]

I know

[queen kernel]

So ?

[calm crane]

i was saying it so you could gauge my computer power
but frankly it's def not enough based on what you said

[dense terrace]

hey, is there a good voice changer i could use without a good gpu?
colab gives me an error when i try upload a voice model

my gpu is terrifyingly horrible i cant run it even in a game

[obtuse quest]

Alguien tiene algun modelo de voz parecido al de young miko o alguna streamer?

[opal ether]

is voice.ai still good?

[minor blade]

Nope, avoid it like the plague.

[opal ether]

o wow ok cause i had it since it released

so i thought it was good

[minor blade]

That program is paywalled garbage.

[opal ether]

yea thats true

[minor blade]

Including the fact it works on the background of your PC, using your resources

Like a cryptominer

[opal ether]

could you tell me how to set up this one? i heard good things only

[opal ether]

btw my name is also Leo so its lowkey Fate

[minor blade]

voice.ai? We don't give support to that program.

[opal ether]

no the one for this discord

[minor blade]

Do you mean w-okada/deiteris fork?

[opal ether]

well the one that i can download voices on

i think its W-okada's

unless theres a new one

#1175430844685484042 im tryna download from here and use it