🧬│ai-chat

[glass junco]

training a better ice cube ill upload in a bout 30 minutes, gonna be better than the last one

[lean tendon]

are you using the ktm 5?

klm i mean

[glass junco]

yeah this one where the mouse is best one

[lean tendon]

So I am getting this exact error

#1327283027776245810 message

this person said to get the latest code,

#1327283027776245810 message

?

what does he mean

[chilly lake]

https://github.com/IAHispano/Applio/archive/refs/heads/main.zip

[lean tendon]

@spring pasture

https://docs.applio.org/applio

Start here

[solar torrent]

W-Okada is the voice changer. For W-Okada, go to #🔍│help-w-okada.

[spring pasture]

thankss

[solar torrent]

Applio is RVC the normal audio converter, not realtime.

[lean tendon]

speaking of applio, I just got a brand new one, and in the preprocess section there is audio cutting

my audio is hundreds of audio files all adding up to around 30 minutes

[lean tendon]

is automatic good? how should I set that section up for audio cutting, chunk length, overlap length

I have a good pc setup so if messing with these will increasing the processing time for quality, I am for that

[minor vault]

ohh

[lean tendon]

nvm I think I got it

#📑│making-models message

My audio is already pre-sliced so I am good

[chilly lake]

depends on the size if your presliced audios

[lean tendon]

my audio ranges from 1-10 seconds

and there a hundreds of them, they are studio quality

[chilly lake]

it would be better to merge them into one file

then trim the silences

then use the new simple slicer so that all slices are uniform

[lean tendon]

ahhhh no, way too much work for me. They are from a very popular game so it is already optimized to minimize silent spaces

halo 2 specifically

i am at the training section

[chilly lake]

then use automatic

[lean tendon]

does "save only latest" refer to saved epochs?

[chilly lake]

g/d weights

if you feel lucky, use only the latest

[lean tendon]

I am not really understanding, so is it beneficial to have both "save every weights" and "save only latest" on? I want a backup incase something goes wrong, but I don't want to rack up storage spacce

[chilly lake]

save every weight - saves the final model every x epoch together with d/g, otherwise saves only d/g

save only latesd - only keeps one copy d/g, no way to roll back to some previous save

[lean tendon]

ok

ah screw it, I'll do every weight

[chilly lake]

every weight + only latest

it is fine

[lean tendon]

👍

[chilly lake]

first model gonna be a garbage one anyway 🙂

[lean tendon]

naaaah

I trained my voice model on this-> https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI

and it came out almost perfect, it was so accurate, and now I found out about this friggen pre trained klm 5 thingy and applio

holy hell, I am excited to see what this will produce

I wish I can upload audio here I want to show you how good the previous one came out

@chilly lake https://jumpshare.com/s/YTWhXW6VDYnfGkQnFfYM

here listen to that, thats my own voice, the previous software got the character sooo accurate

AAAAH

hell

The parameters of the pretrain model such as the sample rate or architecture do not match the selected model.

all that set up and applio again gives me that error

-____________-

maaan this is so dissappointing

I really wanted to use the new klm 5

[tough moss]

anyone got a good STT or TTS model?

[tough moss]

I'd like to do the same. How much data and what microphone did you use?

[lean tendon]

luckily for me, I had a little 30 dollar mic sitting in the back of my closet, I used that to yell at the top of my lungs in a very quiet space like in my car 100 meters away from any of my neighbors in the middle of night

and I used 20 minutes of studio quality dialogue audio from a video game

[tough moss]

could you send me where you got the dialogue audio from video games?

i'd like to make my one personal as well

[lean tendon]

yes, yes

look here

[tough moss]

also is RVC or Tortoise better?

[covert lake]

Those are 2 different things

Rvc is STS

[lean tendon]

@tough moss https://www.sounds-resource.com/xbox/

look here

[tough moss]

ohh

thanks

[tough moss]

what about Tortoise?

sorry, I'm new to this stuff

[covert lake]

Tortoise is TTS

[covert lake]

Dw

[tough moss]

oh

[covert lake]

RVC is made for STS natively, you can use it for tts only if you make a tts audio with another ai and use it as an input

[tough moss]

ohh okay

then could I do Tortoise into RVC?

[covert lake]

Well yeah, tho tortoise is pretty slow and old

[tough moss]

oh I see

[edgy bloom]

Congratulations mok!

Your Froakie is now level 9!

[tough moss]

any recommendations you have?

[glass junco]

bro so crazy at 130 epochs

[covert lake]

Well depends what you're looking for, https://docs.aihub.gg/tts/tts-tools/ can help

[tough moss]

thanks

[lean tendon]

hmm

whats the difference between klm 5.0 and klm 4.4

isn't klm 5.0 better?

[woeful leaf]

hey guys

anyone know where i can download deepseek r1 ?

im on the github but see no links to the file anywhere

is command line the only way to download it ?

i want to have the actual file

[dapper drum]

The main difference between bv1 and the one on X minutes bv2

[lean tendon]

when using a pretrained model for training, does the training data have to be the same sample rate as the pretrain model

for example, klm 5.0 is 44 khz

my training data is 48 khz

[covert lake]

I used ollama tbh

[tough moss]

i've tried fiddling around with some models in the link, do you know which translates the fastest?

[woeful leaf]

its all encrypted tho

[covert lake]

huh?

[woeful leaf]

like i dont think i can copy and paste that to back it up to another system

[tough moss]

should've worded it better
do you know which one does TTS the fastest?

[covert lake]

the links I sent are about TTS, they aren’t translators

[tough moss]

yeah

bad wording on my part

[woeful leaf]

do you know anywhere where i can get the actual file tho ?

i want to be able to back it up

comman line is the only way to download deep seek ?

anyone know another way ?

[covert lake]

well, the fastest tts would be pipertts, it’s not added since it’s lower quality, like 24khz

A better one would be edge tts (cloud), for locally xtts2 or fish speech, not sure much about their speed tho

[covert lake]

it’s not, it’s open source

[tough moss]

maybe you could try to find the file after you downloaded it via cmd?

[covert lake]

the code ain’t obfuscated

[ancient swan]

lm studio

[covert lake]

not open source

[tough moss]

I'm currently hosting it on a server at my IDC.
seems xtts2 or fish speech would give good results?

[ancient swan]

what's not open source

[woeful leaf]

ollama is so weird and like secretive, it doesnt even say where it installs to and gives no information about it

[covert lake]

lm studio is closed sourced

[tough moss]

im currently using kokoro and espeak. do you think they'll give better results?

[covert lake]

only their cli is open sourced

[woeful leaf]

how do i go about doing that ?

[covert lake]

idk what’s “idc”, but the ones I said are 0shot, meaning no training would be required and still can give good results

[ancient swan]

does that matter lmao

you can still download and use it

[tough moss]

I personally have mine installed in a docker so I can find them with enough trial and error
@covert lake Internet Datacentre

[tough moss]

do you know how they compare to kokoro & espeak?

[woeful leaf]

when i found where the model was downloaded to, its not a single file but a bunch of non human readable computer language

a whole mess of folders with random letters and numbers

[tough moss]

probably compiled then

[woeful leaf]

thats how ollama downloaded it

[covert lake]

Smh

[ancient swan]

@woeful leaf you can download lm studio and just search for distilled deepseek r1 models

[woeful leaf]

oh how can i go about installing it the docker way ?

is it hard ?

[ancient swan]

it'll show many variants

[tough moss]

@covert lake do you have any audio files generated by xtts2 or fish speech? I want to see if they're good

[ancient swan]

even 671b variant, if you actually have a system that can run it

[woeful leaf]

oh nice thanks broskie

lm studio ill look it up

[tough moss]

that's what I've currently got going on

[covert lake]

I’ve also heard of lm studio being slower on reddit but i can’t be sure on that, I would prefer something being open source especially for AI

[tough moss]

ohhh.... mb I probably should've recommended that

[covert lake]

are you talking about the .exe ?

[woeful leaf]

oh damn it is ?

[tough moss]

like anything llm or open webui?

[woeful leaf]

shitt

[covert lake]

idk about kokoro & espeak tbh, never heard of them before

[tough moss]

im not sure but i've heard good things about anything llm and open webui

[tough moss]

then i hope xtts2 and fish speech do better then

[edgy bloom]

Congratulations mok!

Your Froakie is now level 10!

New move!

Your Froakie can now learn Lick!

‎

‎

‎

‎

[tough moss]

time to read documentation

[woeful leaf]

your good i just apprreciate yall helping

[tough moss]

aw man

[woeful leaf]

no, but i doubt its an exe file anyways. maybe im wrong

[tough moss]

oh fish speech has a website

[covert lake]

check their github, they only open source thier cli, not the rest https://www.reddit.com/r/LocalLLaMA/comments/1cvawmz/comment/l4o6l5l/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button

[ancient swan]

what, he just wants to get deepseek r1 model and use it, anyone can use ollama or lm studio to use the model, cus the model is open source, it doesn't matter if the apps are closed source lmao

[tough moss]

WOAH Fish is good

[woeful leaf]

but like i just wanna be able to back it up in case something happens to my computer

[covert lake]

his choice what he wants to use anyways

[tough moss]

@covert lake sorry for ping but does fish have any pretrained models I can run locally?

[covert lake]

https://github.com/fishaudio/fish-speech

it basically uses a model that can be used for zero-shot

[tough moss]

oh okay

[covert lake]

the speed will depend on ur gpu ofc

[tough moss]

2x nvidia quatro 5000s

should be plenty

[woeful leaf]

well i have it running on ollama already, but ollama is weird and sketchy and gives you no control over anything and basically hides the files

[tough moss]

edge tts and fish seem to be really good though

[woeful leaf]

i want to be able to back it up on another computer

and use it to reinstall if i decide to reinstall windows

[ancient swan]

@woeful leaf you can use any open source model in ollama or lm studio, i don't know why nick started talking about closed or open source, cus that just doesn't matter, lm studio has more configurable settings, ollama is just load the model and run it type of app

[covert lake]

So, are you trying to find where ollama download the models?

[woeful leaf]

because if i lose my windows and deepseek gets banned for some reason, then that is a problem

[woeful leaf]

no i agree with him i want it to be open source

i value my data and privacy

but other options are still good to know tho

[covert lake]

people prefer open source for transparency

[covert lake]

that’s true

[covert lake]

try looking in C:\Users<username>.ollama\models

I’m not home but the models should be stored here

[tough moss]

you may enjoy anything llm

[polar flax]

^either this, or you can change where the models are downloaded by adding environment variable OLLAMA_MODELS with the path as the value

[tough moss]

last i checked they were open source

[woeful leaf]

yeah thats where i ended up finding it after alot of digging

but it looks like some ecrypted type business

[ancient swan]

lm studio works fully locally once you downloaded the models

[woeful leaf]

non human readable jargan

[covert lake]

You got a point that the models are open source anyways, sorry if I was being rude on that, I was just pointing out that why not use also an open source program since transparency can be important to people

[woeful leaf]

and if im right and it is encrypted then it wont work on another computer which is a problem

[covert lake]

Do you see smt like .gguf files?

[woeful leaf]

so if deep seek takes it down and something happens to my computer then thats a problem

[covert lake]

becuase it’s the type of format the files should be in

[woeful leaf]

let me check again

[covert lake]

thx for pointing out

[lean tendon]

@ancient swan hi Im kinda new to training models and I have been stuck on trying to finally train my voice model with the pretrained model KLM 5.0 with the new applio for a few hours now, do have some time to help me figure out what I am doing wrong?

[woeful leaf]

thanks for the input, were you able to locate the model files or are they kind of jumbled like they are in ollama ?

[ancient swan]

it has models manager

and even shows models location on your pc

[covert lake]

I don’t understnad if you’re trying to read inside .gguf files ?

[woeful leaf]

sha256-96c415656d377afbff962f6cdb2394ab092ccbcbaab4b82525bc4ca800fe8a49

this is the name of one of many files

[ancient swan]

you just choose the vocoder that it was trained on, the sample rate and then choose the pretrains themselves

[woeful leaf]

its not a single file but many folders containing the 4gb that makes up the model

[tough moss]

@covert lake apologies as this is my first time trying out 0-shots.
how long do I record my audio for? should I aim to minimize any background noise, or is there a way to filter it out?

[lean tendon]

yeah, I set it it applio-refinegan, 44.1khz and the pretrains themselves and constantly just pops out this ->

"The parameters of the pretrain model such as the sample rate or architecture do not match the selected model."

[woeful leaf]

my bad turns out its actually one folder but many files with similar jumbled names like that

[ancient swan]

you have latest applio from github, not the precompiled right?

[lean tendon]

yes, I built it myself

i pretty much just ran both .bat scripts and went straight to the training tab

[covert lake]

10 to 30 secs high quality, possibly minimalizing noises

[ancient swan]

yeah, the refinegan has been updated multiple times already, old test pretrains will not work anymore

[lean tendon]

my own traning data is 32 khz, I don't know if that matters

[tough moss]

alright thanks

[ancient swan]

wait until we release new pretrains

[lean tendon]

hmmmmmmm, maybe if I downgrade applio, this could work

[ancient swan]

that might work

[tough moss]

ive got 3 minutes total. is that fine?

[lean tendon]

I don't think that will work actually, something else is wrong, the author said in his post his newest pretrained model works with the latest version of applio from the 30th, "The XE4 version can be used with the newly patched RefineGAN from 30th, January"

[tough moss]

does fish speech support mkv

[ancient swan]

maybe you have an older version then?

[covert lake]

it seems like they use modelfile files for better efficiency, divding the model in multiple layer files https://www.restack.io/p/ollama-answer-model-file-examples-cat-ai

https://github.com/ollama/ollama/blob/main/docs/modelfile.md

if u ever need to move it in another pc or copy it somewhere else, u would be fine by just copying the folder related to that specific model sha

[covert lake]

Btw are u in the usa?

[lean tendon]

nope, its the newest

[woeful leaf]

oh damn never heard of that file type im gonna look into it

yeah im in the usa

[lean tendon]

bloody hell I have some wierd problem

[woeful leaf]

problem is when you download multiple models all the files are mixed in with each other

so i cant single it out and say "this is the model file" cuz its broken up into multiple

[covert lake]

In Ollama, a modelfile refers to a configuration file that defines the blueprint to create and share models with Ollama.

The modelfile contains information such as,

Base Model Reference. All modefiles must have a model that they use as the basis for any new model
Parameters. These specify things such as the temperature, top_k and top_p that should be applied to the new model. We’ll talk more about these later on.
Template. This specifies what the final prompt will be that’s passed to the LLM.
System. We can use this command to determine how the system behaves overall.

This is what I found

hopefully it’s clear enough

iirc each model has an sha hash folder, and in each you can find the multiple layers divided

[covert lake]

you can check what is the sha hash of the model with a command has explained in here

I will check this out when I get home myself, I think you can also get the .gguf files from those modelfile files

[tough moss]

do I have to pay to use fish speech?

[covert lake]

nope you can use it locally

there’s a paid subscription only if u use their site

iirc they make so the newest model is closed source, then when they make a newer one, they open source it too

so its a cycle of them making new models, and releasing the previous model when they make a new one

[tough moss]

am I able to run it on jupyter or must I run a new docker container?

[tough moss]

do you happen to have the documentation for that? I can't seem to find it

[covert lake]

https://speech.fish.audio/#macos-setup

btw u don’t need to finetune their modoes if u don’t want to, u can just use thier models

[tough moss]

im windows

[edgy bloom]

Congratulations mok!

Your Froakie is now level 11!

[tough moss]

well ill try it first

[covert lake]

It’s above https://speech.fish.audio/#windows-setup

[tough moss]

alr ty

[woeful leaf]

oh okay so as long as i have those i can copy and paste it to another computer and have it work ?

[covert lake]

I haven’t tested myself but yeah that’s how it should work like that, https://github.com/ollama/ollama/issues/2860#issuecomment-1976011228

[plain cove]

@stark scarab Gay 0-0

[tepid basin]

@covert lake best ai image generator local for 1650? for @oak fractal

[ancient swan]

sd 1.5

[tepid basin]

needs loras for making characters and shit

[ancient swan]

1650 has only 3gb right

[covert lake]

Prob can't do more than sd1.5 or sdxl

[oak fractal]

Mines 4gb

[ancient swan]

maaaaaaaaybe you'll be able to run sdxl

[oak fractal]

So i have to install the models myself later on??

Dunno much about sd

[polar flax]

no it is 4 GB

[oak fractal]

I just got started

[plain cove]

https://cdn.discordapp.com/attachments/303556951672094722/1326854840097308684/awoouf_0_20250109_082808.gif

[tough moss]

which model has the most up-to-date data?

[polar flax]

[prisma frigate]

@finite cedar can't pm now ask here

how to send images

@finite cedar burh I was hacked

have sense dude

[finite cedar]

I can't know for sure it's you

also the messages that I sent you wasn't directed to you

to the hacker

[prisma frigate]

burh

[finite cedar]

lmao

[prisma frigate]

anyway I changed my subname fo rnow

limited till some time

let see

[finite cedar]

imma un-add you bruh XD

add me when you solve dat shit

[prisma frigate]

is subname good enough hint

nah its fine now

gmail was compromised

[finite cedar]

you have 2fa enabled?

i doubt it

[prisma frigate]

doin it now

[finite cedar]

stoooopid lmao

issokay bruh it happens

[prisma frigate]

sure btw any girl voice model you made?

[prisma frigate]

was waiting for that web girl voice model

her voice is good

[finite cedar]

yeah I stopped making voice models because I reinstalled windows

so I need to re-learn making voice models again XD

[frigid marsh]

just great, rvc collab feature extraction doesnt work and mainline doesnt tell me anything about what happened with my index

i barely do models and when i finished one yesterday suddenly rvc collab doesnt work

man im so mad rn

[covert lake]

Google colabs have issues since 2 weeks #📰│dev-updates

[covert lake]

Send what colab link and what issue u are having in #✨│ai-help

There are fixed colab

If you're talking about rvc disconnected, it's old and broken rn

Mainline colab (from Hina) works

[frigid marsh]

mainline collab for some reason doesnt tell me anything about the index, i dont know what i did wrong

[solar torrent]

Notebooks on Google Colab have been affected by when Google changed the main Python version to 3.11, which made them broken to work.

[frigid marsh]

[covert lake]

Wdym? Elaborate your issue in #✨│ai-help

[wanton plume]

holaaa

[kindred kelp]

Anyone have any code ideas/ suggestions I can implement?

Cool. I have a deepseek ollama implementation already as well

[oak radish]

Nice

Follow me on medium

I will update the next project

what you have to d

do

[digital burrow]

is it just me or weights ai is glitchjing rn

the website

[buoyant oxide]

guys what are some funny voice modules? there are so many of them I can't pick

[lucid thistle]

hi guys, I made an interesting nonprofit fun ai project. I made a short video about it. Asking permission from admins to share.

[grim pewter]

how do i download the voices? when i download them i get a zip file to extract and when i do everything inside dissapears

[pale rune]

Hey guys I'm new to using AI further than gpt but I was wondering if I could make my own gpt like service (for personal use and maybe friends) where I just pay for the API access costs and run it through gui like webchat but a bit more customisable to add things like web browse and whisper.

Seems like it's better paying a dollar per million tokens than waiting for deepseek chat to work or paying $200 a month for gpt pro

[leaden grail]

guys what are some funny voice modules? there are so many of them I can't pick

Hello, I need to level up to access voice models, so I’m chatting here

[desert arch]

eerr

[covert lake]

use #1159290752195633273

[covert lake]

You actually don't need to

U can use #1175430844685484042

[covert lake]

Modules?

[haughty mortar]

Have you tried mixing the voice around in Audacity? You can get some really funny voices from just a normal voice that way - and audacity is FREE!

[glass junco]

chat my 6 minute data set with ice cube go crazy

[rustic meadow]

hola

como busco los moelos

[chilly lake]

#1175430844685484042 https://letmegooglethat.com/?q=spongebob+rvc+voice

[rain blaze]

hello

[rustic meadow]

nesecitola de bad bunny no me aprese

[honest scaffold]

Does anyone know when the collab will return?

[rocky iron]

help

:sad:

my voice changer is not working

i placed in everything the instructions said

my voice changer crashed when i pressed "start voicechanger"

T-T

[kindred kelp]

What should I use my deepseek server for?

Besides undercutting competition for a quick steal

Rather implement a sustainable and worthwhile project

[rocky iron]

bro i actually need help

does anyone konw how to help me T-T

Can anyone actually help me??

pleaseee

T-T

[tough moss]

does anyone have any idea on how to record microphone via only a jupyter server hosted elsewhere (not on the same device) which is connected to via an ip? or do I have to run a local script

[covert lake]

explain ur pc gpu and tut link u followed in #🔍│help-w-okada

[sage canopy]

[solar torrent]

I've done that.

[onyx summit]

https://colab.research.google.com/github/hinabl/AICoverGen-Colab/blob/main/Hina_Mod_AICoverGen_colab.ipynb
Here's the link I used to make ai covers

[covert lake]

that one got a fix recently, retry now

[onyx summit]

Still the same🤕

[covert lake]

refresh the colab and be sure to be using a new runtime

[onyx summit]

I tried it but still the same. Idk why it's keep appearing when I try to generate it.

[covert lake]

@molten stone u sure it’s fixed

[molten stone]

ye fixed yesterday and tried multiple times

the other guy that was having problems also said its fixed

that kinda looks like you either dont have a model or song loaded

[soft flame]

Man why can't i get Audios from YouTube Directly

It was so fun to mess with

[covert lake]

you mean on weights.gg?

[soft flame]

Yea

[covert lake]

you can just download it urself and upload it on weights

youtube blocked the site from downloading iirc

[soft flame]

Well fair enough

[covert lake]

just use ytdlp or any site that comes from googling “youtube mp4 download” like cobalt tools

[soft flame]

But in most cases, some sites just doesn't work for downloading

[covert lake]

use cobalt tools or https://yt.savetube.me/

[soft flame]

Ok

[covert lake]

you just need to google some sites that’s all

[onyx summit]

I do everything correctly and can you tell me should I insert a text in the token? Also I paste YT link it the song and is it OK? Actually I'm not using this link from the beginning and this link is I used from the beginning https://colab.research.google.com/github/SociallyIneptWeeb/AICoverGen/blob/main/AICoverGen_colab.ipynb#scrollTo=NEglTq6Ya9d0
But it's not working so that's why I'm asking for a new link but everytime I tried to generate with that link (https://colab.research.google.com/github/hinabl/AICoverGen-Colab/blob/main/Hina_Mod_AICoverGen_colab.ipynb) it's not working

[odd knot]

i use yt1s.com for saving vids

bc

well

yeah

as for saving audios i use cobalt

tho i could save videos on cobalt but it wouldn't be too compatible with other softwares

[verbal minnow]

hungry

[timber quartz]

albay9795

Avatar Link

Requested by makidanyee

[civic trail]

Oh

[tough moss]

is openai's text to speech free? I know that whisper is free but what about the other way around.

[covert lake]

don't think so

we are english only, unless you use a channel specifically made for another language

[tough moss]

alr. that's fine. could I ask for some advice?

I want to talk to an AI hosted on another server.

should I make the voice recording client or server side?

i.e. client with a .py file or website with a recording method.

[covert lake]

i think website with recording method would be easier

[tough moss]

alright. will do that then

[elder willow]

Self-mackerel profile

yo im new to this training shit, how do i know when my model starts overtraining?

[night lake]

https://docs.aihub.gg/rvc/resources/training/#usage-guide

[pallid lily]

Will deep seek affect RVC since the technology allows slower hardware to run ai just as fast as better hardware?

[final nimbus]

Today is my birthday 🎉🎉

[clever arrow]

HELLO

its been a while since i been here

does anyone know if wokada is still the best voice changer client? for rvc?

[covert lake]

Wokada deiteris fork is the best, tell me ur PC GPU in #🔍│help-w-okada

[covert lake]

I mean, it's 2 different types of AIs

And RVC runs already on consumer GPU

[clever arrow]

rtx titan im using linux jammy ubuntu

oh my bad in that chat i got u

[hollow tangle]

@covert lake do you have the uvr beta download?

[twilit pond]

ayo @covert lake wassup

ok thats it im convinced your online 24/7

every single time i check the chat theres a message that you sent pretty recently, literally never offline 😭

[covert lake]

I think the latest is https://github.com/Anjok07/ultimatevocalremovergui/releases/download/v5.6/UVR_12_8_24_23_30_BETA_rofo_full_install.exe

[covert lake]

wassup ?

[covert lake]

[hollow tangle]

found it, thanks anyways

[twilit pond]

so like i had a question rq, i assume model makers cannot take commissions at all?? only model masters right?

wanted to make sure

[covert lake]

yup

only model masters can take paid commissions

model makers can only do free requests

[twilit pond]

i see

so like what about requests that are not posted on #1159289738314919936

lets say a guy i met on this server asks me to make a model, can i not take it as a paid commision? is it against the rules

[covert lake]

you can't take paid ones at all

also users should do requests in the specified channels #1159289738314919936 , or contact a model master directly #1191429836321849435

else you can be a model master with /jointeam

[twilit pond]

alr, good to know

[balmy vector]

Is there a Jane Doe from zzz voice model?

[covert lake]

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://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/

[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

[vestal shadow]

Sorry if this isn't the right place to ask this but hopefully it will be okay as it is about voices.
Has anyone ever come across the app DubbingAI?
https://www.dubbingai.io
I'm just wondering if you have whether you think they are using RVC somehow inside the app or do you think it's something custom coded?
Their voices do sound very good indeed and all of a very high quality so wondering if it's RVC or something else?
Also if anyone is a developer do you have any idea how much it would cost for a developer to code an app like that? Thank you.

[kindred kelp]

Is there something wrong with the API for o3? I can't access the new model released yesterday from the api

[shrewd aurora]

https://youtu.be/cbMccIg9sLk?si=suqyAXKXWPE2SBzS&t=115 Is there any term for voices speaking perfectly in sync like this, or any models someone knows of that does it

[steady surge]

anyone have a guide to set up deepseek locally?

thats up to date

[hollow tangle]

just use ollama and cli

[steady surge]

what model would you recommend?

[hollow tangle]

what gpu and how much ram you got

[steady surge]

4070 12gb with 32gb ram

[hollow tangle]

14B

[steady surge]

how does that compare to 32b?

[hollow tangle]

keep in mind local deepseek is pretty stupid

like every local llm

32B tho compares to o1 mini

[steady surge]

how bad would 32 run?

[hollow tangle]

very slow

[steady surge]

imma just download both and use 14 ig

[hollow tangle]

i have the problem that ollama uses cpu instead of gpu

[hollow tangle]

yea just try, wont hurt

[steady surge]

since a bill was just introduced to make it a 20 year prison sentence for downloading AI from china

[hollow tangle]

wtf where

[steady surge]

sec

heres a thread https://xcancel.com/opensauceAI/status/1885483639611531704#m

[hollow tangle]

the us??

[steady surge]

Yes

[hollow tangle]

tf

[steady surge]

if you havent noticed theres currently an internal coup

[hollow tangle]

trump?

[steady surge]

https://www.hawley.senate.gov/wp-content/uploads/2025/01/Hawley-Decoupling-Americas-Artificial-Intelligence-Capabilities-from-China-Act.pdf

[steady surge]

Elon and Trump

[hollow tangle]

guh

[steady surge]

Elon was given full, unrestricted access to the US Treasury earlier today

[hollow tangle]

im not in the mood to intoxicate myself with us politics 😭

[steady surge]

including records of all social security numbers, addresses, 6 trillion in transactions, etc

[steady surge]

do you live in the US?

[hollow tangle]

no

[steady surge]

lucky

[hollow tangle]

thank god i dont

[steady surge]

when it gets too bad im just gunna kill myself

idc anymore

[hollow tangle]

actually dont dude wtf

[steady surge]

if theres a god hes a spiteful piece of shit

[steady surge]

better than being put in a concentration camp

[hollow tangle]

are you trans or an illegal?

[steady surge]

Theyll come for me eventually

ive donated to certain political parties

[hollow tangle]

okay but for what

[steady surge]

pretty much all prices are going up my 25% in 2 days too

[hollow tangle]

but eggs go lower!

[steady surge]

thats why grocery stores are starting to impose limits on how many eggs we can buy!

[hollow tangle]

lmfao insane

[hollow tangle]

not even during covid

we need more chickens 😔

[steady surge]

why cant i tell ollama where to install

[hollow tangle]

wdym

[steady surge]

it installs in appdata without giving an option

.fm

[indigo pike]

Now playing - Pandorify

The Rest for the Wicked

Sohodolls • Ribbed Music for the Numb Generation

41219 total scrobbles · 🕓 4:01

[steady surge]

when ollama is installed, how do you like

interact with it

[hollow tangle]

cmd

[steady surge]

the app is open but it doesnt open a cmd, and if i open a cmd the ollama command isnt recognized

do i have to run the cmd in the ollama directory?

[hollow tangle]

no, the normal cmd should work fine

strange it doesnt work

[steady surge]

imma reinstall

does ollama auto update models or do you have to manually pull them

[hollow tangle]

idr

[steady surge]

id rather not get 20 years in prison if it does

[hollow tangle]

nah you wont

[steady surge]

you doubt the fascists in office

[hollow tangle]

yes

[woeful leaf]

i know right ? i thought that was weird too. its all secretive. i guess not that suprising tho after all its by meta

[steady surge]

ew

fascebook

.fm

[indigo pike]

Now playing - Pandorify

Money

Jazz Emu • Technically, Probably More Of The Same, I Suppose, In Theory

41235 total scrobbles · 🕒 3:22

[woeful leaf]

yup. its terrible.

anyone know any open source ways i can use deep seek thats not ollama ?

[hollow tangle]

why

[woeful leaf]

cuz ollama is from meta

and meta is on some creepo weirdo shit

not down to use anything from meta

dont trust them for shit

[rich sable]

Hello! I just watched this video on creating an app with deepseek, mobbin, and cursor ai: https://www.youtube.com/watch?v=WZ8g6deOyAk&ab_channel=BuildWithAI. So it is an amazing tutorial but I have one question, I am a complete beginner with app development. At timestamp 8:34, he said that you worked on the app for a while and showed us a basic result. I have no idea how he got there from what you had before, did he just prompt the ai to continue, if not could anyone please provide more detail? Thanks.

[lean tendon]

does overtraining ever worsen a voice model?

I am basically going to let a model train overnight

[steady surge]

https://youtu.be/wox6w4PeKpE

[dapper drum]

Can someone tell me how I exactly activate the kaggle AI cover maker

[kindred kelp]

Manually pull. And you can access it using cmd line. I use putty on my Linux cloud server, it runs faster on a good server anyway compared to my local machine

[steady surge]

I don't want it to auto update so that's good

[steady surge]

because

[kindred kelp]

Use the deepseek API or download the models from huggingface repo and make your own chat interface

The model files can be downloaded without ollama but ollama is pretty good for my purposes - I just tunnel into the command line and request it using JavaScript then send to my client from the server that's running it

[kindred kelp]

Is this real?

[steady surge]

the bill was introduced

[kindred kelp]

I'm so glad I downloaded my copy weeks ago.

When

[steady surge]

sometime in the past few days

[kindred kelp]

Great. I'm happy personally because I already have a working copy

This eliminates any competitors bwuhahahahaha

I'm fricken ecstatic

I'm glad I was so diligent lately with my coding routine. Good job congress you made me rich

They did that stuff just so competitors wouldn't appear.

Openai grok and Google lobbyists no doubt

They were all standing at the inauguration. This is terrific news

For me.

How in the hell is that bill even valid freedom of speech....... Unless you speak Chinese a.i. model

[steady surge]

it's a bill

it hasn't passed yet

[polar flax]

it is foss, another one is LM studio which is also free but closed source

[kindred kelp]

I hope it passes

F yeah hard work and dedication paying off

[polar flax]

if not sure, try add the ollama path to the PATH environment variable

[edgy bloom]

Congratulations chibi hater!

Your Gengar is now level 36!

New move!

Your Gengar can now learn Sucker Punch!

‎

‎

‎

‎

[kindred kelp]

Bill will be enacted in 180 days. You guys should, get a working copy while you can. But if you don't that's cool with me too cause then you can pay me to use it

🙂

I'm so related

Elated

[covert lake]

It's just not everyone got a 1tb of ram to run the 761b one

[covert lake]

Fucking racism

[woeful leaf]

damn looking for open source only, still good to know tho

[covert lake]

It could

[covert lake]

Be sure u open ollama while using the CMD

Ollama has to be opened in the background

[covert lake]

Ollama ain't from meta btw

[woeful leaf]

whats an api ? or how do i go about making my own chat interface with the hugging face files ?

[covert lake]

Application programming interface, an example is u pay for ChatGPT API and they let u use their ChatGPT via code in your programs

[kindred kelp]

An API is a service that allows you to interface with code on a server. It has endpoints that you can access for information. You can't legally in the United States as of yesterday

It faces a 20 year prison sentence

[kindred kelp]

Well it's pending. 180 days

[polar flax]

well it's not the meta's llama models themselves

[kindred kelp]

I don't know for certain you can still legally do it

Get ollama and run a simple command. I just don't advise you to break any laws in the meantime

[covert lake]

I hope it's a fucking joke

Cus it's just racism now

[stone lava]

omg rip rvcdisconnected 💔💔💔

[covert lake]

Yeah it's discontinued forever

Rip

U gotta adapt to a newer colab

[stone lava]

i used to use tht like a lot

[stone lava]

yea im using kaggle rn

[covert lake]

I used that back in the days

[stone lava]

rvcdisconnected will be missed

[kindred kelp]

I'm using exocode

[woeful leaf]

ohhh okay so it basically lets you plug the ai model into other programs ? do i have that right ?

[kindred kelp]

Kinda. You use an API to "plug into" software that provides you with a service or access to functionality

[covert lake]

It's paid and Im not sure if that will be banned too tbh

[woeful leaf]

oohhh okay thats good to know. damn seriously ?? so illegal to use now ? damn i knew they were gonna ban this shit

[covert lake]

^

[kindred kelp]

You can have a payment gateway API or a real estate API

[covert lake]

They can ban whatever they want, but they can't touch an SSD

[kindred kelp]

Solid state drives?

[woeful leaf]

bro trump is openly racist why are you barely noticing this lmao thats nothing compared to the other stuff

[covert lake]

Yeah I meant as if u download it they can't remove it from u

[kindred kelp]

Right
I have deepseek on a flash drive

They can arrest you if you download it after the expiration date

[woeful leaf]

oh damn thanks for the lesson fam

[covert lake]

How would they know

[snow sedge]

There's nothing illegal in API as long as you didn't discover it through reverse engineering and if it doesn't provide data you're not allowed to legally access (PII data that's protected by GDPR and stuff) or expose to others

[kindred kelp]

Dunno

Emojikage, not talking about that. The API question was a separate conversation

[covert lake]

We were talking about USA banning deepseek

If they ban it they prob gonna ban the API too

[kindred kelp]

Dunno

[covert lake]

We will see ig

VPNs still exists btw

[kindred kelp]

But all the tech giants were at the presidential inauguration and they definitely want less a.i competitors coming out

[snow sedge]

It's not illegal to access blocked websites/their APIs either

As long as they don't provide something illegal

[covert lake]

I meant they would make so u can't access the API maybe

[kindred kelp]

They banned tiktok for a day or so

It's very possible they can

There's precedence

[polar flax]

I still feel that they just hate the competition

[snow sedge]

Then you'll see a massive spike in VPN installations

[covert lake]

Fuck yes 🔥

[snow sedge]

Or someone will just make a frontend for Deepseek LLM that's accessible to anyone

[covert lake]

true lol

U can't ban things off the internet this easily

[covert lake]

Yup

It prob will be also in the internet archive and dark web anyways

[snow sedge]

In Russia, they block access to ChatGPT and OpenAI blocks registrations with Russian phone numbers. But who cares when you can register an account with different phone number and access it with VPN

And people buy ChatGPT accounts with Plus subscription and share. Who cares as long as you pay

[covert lake]

Just the same shit when other apps get banned (yuzu)

[kindred kelp]

I'm making a code editor called exocode that uses deepseek. To be determined price but it will be less than everyone else for sure

[kindred kelp]

Working on it

[woeful leaf]

nicee let us know when its done cooking

[kindred kelp]

For sure
Thank you

[woeful leaf]

yeah its basically another emultator that got taken down thats got like elleventy other sites you can download it from

has anybody tried the jailbroken version yet ?

[twin pulsar]

anyone know any good optimized settings for okada vc

[solar torrent]

For W-Okada, go to #🔍│help-w-okada

[covert lake]

use #🔍│help-w-okada and show me a screenshot of ur wokads

[twin pulsar]

alr thx

[solar torrent]

No.

[ionic pumice]

GAMBLING 🤑🤑🤑🤑

[solar torrent]

Money on my mind. 🤑

[worthy coyote]

Money Money In My Mind

Let's Go Gambling!!

[polar flax]

for the chance to win a 5090 purchase, but it's joever already

[wise dune]

peace une

[timid mulch]

Hello who cant help me i want a voice girl clean

[covert lake]

like model or program?

[timid mulch]

model and settings

[covert lake]

show a screenshot of ur wokada in #🔍│help-w-okada

[lean tendon]

i just trained my data set on KLM 5.0, Titan, and Ov2Super on Applio and they all came out trash

I used the same exact dataset, same settings with RVC-Project off of github

its 10x better with the built in v2 pretrains, whats up with applio?

[minor blade]

On my honest opinion, i only recommend sticking with OG pretrain.

[lean tendon]

I think so too

[minor blade]

Yep, all others got issues the OG one simply doesn't have.

[lean tendon]

I got really hyped learning about all these pretrains after training my first model

sucks to be dissappointed, but hey I am glad I found something that works

btw

you want to hear some sweeeeet voice lines from the og? I voiced them all myself

[lean tendon]

I guess its a case of speciality, singing voices, other languages, anime

and og is the more versatile one, my characters and the ones I plan to do are more wierd sounding so the og seems to be a good fit

[lean tendon]

thanks for confirming my thought, I been stuck on looking for the best pretrain for a few days

[minor blade]

These are voicelines you made tested with your models?

[lean tendon]

correctamundo!

[minor blade]

I see.

Altho for sake of safety don't post links of these (I'm not sure, just for sake of security)

[lean tendon]

unfortunately I can't post audio here so 🤷‍♂️

[uneven inlet]

hola me pueden ayudar? quisiera saber como poner para buscar un modelo de voz?

[minor blade]

Mijo, los modelos se buscan en el canal de #1175430844685484042

[uneven inlet]

de verdad hermano, pense que era en find models.. estoy buscando uno para narracion estilo morgan freeman o en español conoces alguno?

[minor blade]

No bro, no conozco ninguno.

[uneven inlet]

o alguna voz de narrador amigo, se puede usar modelos de ingles para narrar en español?

[uneven inlet]

amigo esta es la unica alternativa de applio o me equivoco?

[minor blade]

Si usas samples en español para testear con modelos entrenados en ingles, por supuesto que si

[minor blade]

Si, AI HUB/Weights son la unica alternativa confiable a Applio.

[uneven inlet]

gracias hermano de verdad, un abrazo soy nuevo espero aprender como usarlo y crear voz propia

[tired aurora]

saludos, alguien podria ayudarme a clonar la voz de georgina rodriguez

[minor blade]

Si quieres puedes hacer un post en el canal de #1159289738314919936 pidiendo ese modelo. (puede ser gratis o pago, aunque pago/paid te da más chances de que los model masters te contesten)

O alternativamente puedes leer las guias y aprender tu mismo a hacerlo.

[dusky grove]

It only let's you make 1

[kindred kelp]

Hm I vaguely remember it being pay for generations

You can upgrade

[dusky grove]

Im broke

[kindred kelp]

Sorry to hear that

[dusky grove]

Can I use your acc?

[kindred kelp]

I don't have one I used it years ago

Sorry

[dusky grove]

K

[kindred kelp]

I would though

[dusky grove]

Really?

[kindred kelp]

Yep if I had one

Sorry

[dusky grove]

Can you make one

[kindred kelp]

No no money either

[dusky grove]

K

[dusky grove]

Do you have any free alternatives

[covert lake]

@dusky grove @kindred kelp play.ht? You can make free speech to speech voice models with RVC

[kindred kelp]

Cool

[covert lake]

Rvc is open source

It's what we all use

[odd knot]

is that lain???

[polar flax]

still better than some hallucinations/blind speculations

[tepid basin]

I don't even understand most of these riddles 😭

I've been seeing these recently lol

[zenith lintel]

AHH

[solar torrent]

Play.ht? Never heard of this site. But I only use Weights.gg for draft AI voice cover, which in turn uses RVC, and it's free.

[foggy kindle]

I have a problem with Rvc, can you help me?

[covert lake]

elaborate ur issue in #✨│ai-help

[foggy kindle]

are you italian, can dm you?

[covert lake]

are you italian too?

[foggy kindle]

yes yes. That's why

[covert lake]

alright, then sure dm me

[foggy kindle]

sorry but, mi dice I can't deliver message to you

@covert lake puoi mandare tu un messaggio per favore, che mi dice che non siamo amici e non mi fa mandare nulla

[covert lake]

alr i will dm

[covert lake]

#1159290752195633273

[iron cliff]

😉

[covert lake]

Use #1159290752195633273

[polar flax]

any point of training your own dereverb model?

[chilly lake]

you take a million of audios, you make a copy of them with reverb added, then you train a model that turns reverbed file back to original

and there you have it

[tender summit]

how do i make an ai model that analyse text and extract some specific info from it

[polar flax]

https://cdn.mos.cms.futurecdn.net/pX9HAHDAM6L6fQ4RiFihfD-970-80.png.webp

[worthy coyote]

reverb remove ai?

is that possible?

if ai can do it, human can do it too

[polar flax]

it's even bazillion amount of data with lot of various reverb plugin effects

[worthy coyote]

just hire a sound engineer

[tender summit]

help

[polar flax]

may not even be enough to work on 99.99% of possible cases

[tender summit]

im tryna cook smthin

[worthy coyote]

nice cook

[tender summit]

btw

where u guys from

[worthy coyote]

hometown

good old days

[sharp yew]

do you guys know any ai cover site? the one I use doesn't work anymore

[ancient swan]

yeah this looks more correct

at least looking at specs 61 vs 82 vs 104 tflops

@stark scarab ban this fucker

[stark scarab]

https://tenor.com/view/anime-sword-art-online-gif-22934382

[minor blade]

Un grande el Eddy

[wanton pawn]

what is faster t4 or 4060

[ancient swan]

4060

[wanton pawn]

even with half the video ram ?

[ancient swan]

vram doesn't affect the speed

[wanton pawn]

sorry i should have specified

training not inference

[ancient swan]

yes correct

raw performance of the 4060 is higher than t4

[wanton pawn]

but the batch size has to be lower no?

[ancient swan]

yeah, but 8 gb should be enough to train at 12 batch size which is maximum you need for small models, also bs isn't as important, as long as it's not 1, it'll sound ok

[rugged plover]

hey, question

anyone got an AI that can help me black and white an imagie while keeping one certain color ?

(please dm me :3)

[orchid crag]

English cuestión:
"Does anyone know of any TTS or audio-to-speech model that can be used from the browser on Android... But that has very realistic audio quality? Not as much as RVC, since replicating that on Android from a random website is not very feasible, but at least I would like to be recommended a site with voices superior to those already offered by Azure or other websites."

Spanish:
Alguien conoce algún modelo tts o audio a voz que se pueda utilizar desde el navegador en Android... Pero que tengan una cálidad muy realista de audio? No tanto como rvc ya que replicar eso en Android desde una página random no es tan factible pero almenos me gustaría que me recomendaran alguna página con voces superiores a las que ofrece ya azure o otras webs

[humble frost]

I downloaded some voices with the ending npy, do you know what these files are?

[polar flax]

only that npy file? you've got scammed, it's like getting an RTX 3090 that turns to be missing its GPU and memory chips, rendering it unusable
https://www.tomshardware.com/pc-components/gpus/nvidia-rtx-3090-sent-in-to-repairer-was-missing-the-gpu-and-all-its-memory-chips

[west gorge]

Anyone got plug ai discord link

[wintry tartan]

thomas shelby?

[polar flax]

ur dad

[latent berry]

a

[warped pagoda]

Hi

[ionic forge]

My student just took a photo of my PowerPoint presentation which is an assignment I gave to them and they upload their photo to ChatGPT to let AI solve my assignment.

I feel in shock

Is there any way to prevent AI scan my assignment???...

😢

[warped pagoda]

would adding a text at the top which should act as a prompt for chatgpt which asks it stop? When it uses text recognition , your instruction is the first thing it reads. may be you can mask it using morse code?

unless they decide to crop it

[polar flax]

you might think they were cheating like if they were using a calculator in a math class, but unlike calculators chatgpt is actually prone to hallucinations

[warped pagoda]

btw bro could you help me with video generation?

[ionic forge]

I feel like I am useless teacher

When I saw that

At this point, Robot can teach every student instead of a real-life teacher now.

I don't wanna be a teacher anymore..

Even tho I am still an intern teacher

but I am disappointed of my student so much.

[pallid lily]

The 5070 has 1000 ai tops and the 4070 has like 3-400 does that mean the 5070 is 3x as fast as the 4070 in ai tasks?

[ancient swan]

These are mostly tensor cores, so if something utilizes them then it'll be faster but not 3 times faster

Maybe like 50% faster, not sure

[polar flax]

given the same vram, I'd prefer the cheaper one

[fervent citrus]

Bro i want real-world based application idea or projects its very urgent...

[tepid basin]

Can you make an app that overlays 100% pokemon coordinates for pokemon go on your screen, and when you tap it, it pushes the coordinates to GPS Joystick for seemless shundo sniping?

you will also need to scrape pokemon go data from their servers

[plush fog]

not sure if there's a dedicated channel for AI music stuff, sorry if posting it in this channel is against any rules!!

[fervent fjord]

lol i am un able to understand any thing here i am a new engineering student

can anyone guide me ???

[covert lake]

pls use #1159290752195633273

[covert lake]

guide u to do what? and what's ur pc gpu

[solar torrent]

Guide of what?

[fervent fjord]

i dont know i am a beginer

[solar torrent]

Please be specific on which you're looking for.

[fervent fjord]

i am 1 year engineering student in computer science in ai and ml bcoz of that i wanted to stay updated on ai related topics but as i am ultimate beginer i dont know anything

[solar torrent]

Oh, lmao.

[fervent fjord]

what should i focus and what should i learn

i am currently learning java which is taught in my university

what should i do?? i want to specialist in one perticular field but i dont know what thing i should specilize and what has demand for job

[solar torrent]

Huh. There are many ways to learn on how to program, unfortunately. But one of the easiest ways anyone can do is playing around with a chatbot like ChatGPT and Gemini.

https://cdn.discordapp.com/emojis/1016021511913603202.webp?size=48

[polar flax]

which AI field are you interested on?

[fervent fjord]

even i dont know what are the diff fields in ai

[solar torrent]

I've been learning in college for four years, and I only know basics teachers taught me.

[fervent fjord]

i am telling uh right i am a ultimate beginer

please tell what is the best field in ai so that i should focus and get the job

do anyone has my solution??

[solar torrent]

"What should I do" with no such of any goal of it is one of the confusing questions I've ever seen.

It's not like everyone knows what you wanna do, unfortunately. For codings, you gotta start off the easiest thing you can do like Python and C. But Python is easier to learn.

That's my suggestion, you may not need to do it like mine.

[fervent fjord]

i what the best field in ai ??

[solar torrent]

[fervent fjord]

lol tell na

as of ur experience what should i learn first

[polar flax]

"what's my best life goal?"

[solar torrent]

I do what I think I can do.

[fervent fjord]

step by step

[solar torrent]

Of course.

[fervent fjord]

i dont know even what are the fields in ai and i am going to become a engineer

great!

i didn't get my solution

Sadly i have wasted my own time for no reason

[solar torrent]

You can't expect anything from people you don't know to guide your goal at every step. So what you think it's easy, you gotta learn and try to do about that subject.

[fervent fjord]

i will figure it out

what is my goal and what should i learn

[solar torrent]

You have selected your way to learn about computer science in university. So just follow what teachers teaching you, and you'll figure it out by yourself.

[polar flax]

you are a sailor who doesn't have a map and compass, and doesn't even know where to go

[fervent fjord]

its not about that i want to bbuild a strong foundation in perticular thing

[solar torrent]

I went on my way to learn about information technology in college, because I'm always aware what my goal is. I didn't have to ask anyone what should I do about that.

[solemn zephyr]

hola

[plain bobcat]

Hello I don't know what is the best place to ask this but I've trained a voice model for kindred, do i upload it to #1175430844685484042 ?

[hollow tangle]

dude are you 7 😭😭😭

[jovial island]

No bruh why is everyone asking me that 😭

[jovial island]

Actually im 15

[hollow tangle]

you should be 5 to belive this shit

[jovial island]

And plus its for fun lol

Stop taking it personal 😭

[hollow tangle]

its not i find it funny you do

[jovial island]

Lol same that's why i sent it

[rustic violet]

m

[kindred kelp]

My deepseek server took 3 minutes and 50 seconds to respond. Guess I should upgrade

I'm working on a code editor that uses deepseek

I call it exocode

[rancid goblet]

@minor blade Sorry for disturbing you, but I did manage to get through it correctly. I also want to utilize the AI voice mod on Discord, but it isn't working properly. Do I need to disable the virtual cabel in audio settings?

woops wrong channel

[polar flax]

#🔍│help-w-okada is the right channel

[minor blade]

Yep, also i'm going to sleep..

[tepid basin]

anyone else using poofartpoo fork?

[kindred kelp]

Anyone see trump talking about 500billion into Stargate? The a.i. infrastructure program

[forest crypt]

gradio down

Cate

[solar torrent]

You may submit your trained voice model to #outdated-model-maker-role. After you submitted it, wait until one or some voice model quality checkers to verify yours and it will appear in #1175430844685484042.

[plain bobcat]

The only problem is that i have it on google drive because i still haven't figured out how to upload it on huggingface

[night lake]

https://docs.aihub.gg/extra/model-maker-role/

[boreal cargo]

yo

[ancient swan]

the what?

[tranquil kite]

Anyone knows where I can find those old models from Applio?

[calm venture]

hello guys, is there an AI capable of see video and describe what it's happening in a video? not using a trascribe function just the video. Like a video of a dog getting on 2 legs and then fly

[woeful geyser]

yo guys there way to tranfar model form weights old account to new one 🙂

[lean tendon]

Sooo whats a good image generation equivalent of rvc-project/applio?

something free

I am thinking I could train an AI based on a particular artists style that I really like, and generate some images

is that considered plagerizing? Can 2d art styles be copyrighted?

[covert lake]

Equivalent to applio? Applio is a fork of RVC, Speech To Speech, has nothing to do with image generation at all

Maybe you confused projects?

[woeful geyser]

hello ?

[tired escarp]

iam looking for a prompt engineer for a claude.ai project (has to do with writing dm me , paying)

[covert lake]

#1159290752195633273

[minor blade]

Use the #1159290752195633273 channel

[open harness]

Check out promo channel 🙂

[solar torrent]

Check deez nuts.

[solar torrent]

I don't pay for text writing. That's too low.

[polar flax]

if the artist disallows AI training as in their portfolio/account page, just don't

[solar torrent]

You can train an image or voice model of anything. If the subject is copyrighted or the author did not allow it to train in AI, just better no.

[tired kite]

Hello, good day! I'm new to this Discord, and I wanted to ask if anyone knows of an AI that can do the following:
I have many videos uploaded to TikTok and YouTube Shorts, and I want to download all those videos with their titles and descriptions in 1080p or 720p quality so I can upload them to my new Facebook page without having to download each video one by one.
Also, it would be great if it could either save all those videos in a folder or directly re-upload them to the Facebook page.
Is there an AI capable of doing this?

[covert lake]

you don’t even need an ai for that

use ytdlp on github

[tough moss]

when accessing AnythingLLM with API, is there any way to remember chat history? or have something similar to GPT's memories?

[tired kite]

Ohh I don't know how to install things from GitHub or how to program them to work. Could you give me a tutorial or recommend a video about it? Thanks!

[tight thicket]

Yay fumo

Fumo fumo fumooooo

Give gumo

[elder willow]

hello

[spark acorn]

Hi

[covert lake]

https://github.com/yt-dlp/yt-dlp

The commands are in the readme

[stable elm]

guys anyone know how i can make my voice laggy?

[covert lake]

what

[stable elm]

make my voice laggy

laggy

you know when your pc burning with 2 fps how is your sound in discord?

[polar flax]

https://tenor.com/view/laptop-smoking-fire-burning-lag-gif-19373925

[stable elm]

glitchy i mean

[steel kite]

guys, do you feels like deep seek (R1) very biased?

[ancient swan]

download cyberpunk and let it be in the background, download some cpu benchmark software and launch it too

[solar torrent]

Imagine trying to run W-Okada on a PC of over ten years old.

[stable elm]

nice idea

[polar flax]

run furmark or cyberpunk with full path tracing in 4K native, no DLSS (dont tell me if you have a 5090 lol)

[stable elm]

my house will explode for that

not just my pc

i have 1998 graphic card with 2m of v ram you think i can get 1 fps in 1080?

[plucky holly]

@icy pendant

Can u help me rq

And go dms

[queen kernel]

Download the mining virus and done 🤣

[whole spear]

ive made a SONG dissing my dad at age 5 by ai wanna hear it

[plain cove]

@whole spear -> #1159290752195633273

[whole spear]

thx

[plain cove]

❤️

[tired kite]

thanks so much ❤️

[covert lake]

Yw

BUT @tired kite WATCH OUT
YouTube be sus about your IP address if u download a shit ton of videos

Since it's technically against their ToS to do that without premium

[lean tendon]

What program do you guys use for generating images and training a model?

I need to create a storyboard

[jolly veldt]

@covert lake just click on code button and choose download, itll prob be a zip file and you unzip it a new folder then open it up and you will see your files

prob be a .exe im guessing

why do the old style artist hate ai so much? way i look at is i cant sing write music or play an instrament, so when it comes to all that im handicapped. so ai just gives the ones that cant do that a way to do it/ all it does is gives many people a chance to make good music. besides most the artist use some type of computer/ai to make their music, and they call that ok but only if your one of them, ever see many rappers use instraments? all computer generated. i just recently got into this and i love it, itf fun. and its not easy like people think i can spend a week just on getting 1 song the way i want it

[drowsy zephyr]

Hello

is UVR7 Real? Someone is telling me it is with no source whatsoever.

[solar torrent]

UVR7 released? I've only heard of UVR5.

[solar torrent]

Why do AI bros can't get over the real artist?

When you let Suno or Udio to generate a track in the style of your favorite artist, it is just the way it is.

[outer apex]

who should i make sing story of undertale

[serene bobcat]

hey just out of curiosity
how quickly can yall get w okada to do its thing on ur hardware
im using a 2070 super and it give me around 0.8 seconds of delay

[solar torrent]

For W-Okada, go to #🔍│help-w-okada. The chat isn't where you ask for how to setup W-Okada.

[solar torrent]

Toby Fox. 🔥

[oak flower]

Text Preprocessing for NLP: A Step-by-Step Guide to Clean Raw Text Data
A Beginner’s Guide to Cleaning and Preparing Text Data for NLP Models + Hands-on with Python
https://medium.com/towards-artificial-intelligence/text-preprocessing-for-nlp-a-step-by-step-guide-to-clean-raw-text-data-2bb8918a4e2c

[solar torrent]

[polar flax]

I have yet to confirm if that article is somehow AI-generated or not

[quartz bay]

does anyone know if I can test a RVC backup audio before it's done fully

im at like 50 epochs but im going for 300 with a dataset of around and hour and a half

i dont use discord a lot so tag me if you have any answers

[serene bobcat]

i swear i gotta be speaking some alien language or something

how Fast have Yall been able to get that shit going for Yall

im tryna measure up with other people, compare specs and speeds
like i.e. some guy with a 4080 can run well with a delay of x seconds

im not tryna recieve tech support

[glacial girder]

Check Privacy. @quartz bay

[elder willow]

Demn

[tawny night]

-colab

[red frost]

📒 Google Colab Notebooks

• Applio, by IA Hispano Google Colab
• RVC Disconnected, by Kit Lemonfoot Google Colab
• RVC Mainline, by Hina Google Colab
• Hina's Mod AICoverGen WebUI, by Hina Google Colab
• AICoverGen-NoWebUI [English], by Ardha, fixed by Eddy, Hina and Gdr Google Colab
• AICoverGen-NoWebUI [Spanish], by Eddy, Hina and Gdr Google Colab
• UVR5 NO UI, by Eddy Google Colab
• UVR5 UI, by Eddy Google Colab
• Hina's Modified Original W-Okada's Realtime Voice Changer, Google Colab
• FaceFusion UI, by Nick088 Google Colab
• FaceFusion NO UI, by Nick088 Google Colab
• EasyGUI, by Rejekts Google Colab
• 🆕 Music Source Separation Training (Inference), by Jarredou & Makidanye Google Colab

ℹ️ Note

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

More commands: -audio, -hf, -local, -kaggle, -realtime, -rvc, /help

[woeful geyser]

-kaggle

[red frost]

📘 Kaggle Notebooks

• Applio Notebook, by Vidal Kaggle
• Applio Notebook, by Shirou Kaggle
• Music Source Separation, by Shirou Kaggle
• UVR5 NO UI, by Eddy Kaggle
• Original W-Okada's Voice Changer, Kaggle
• Modified W-Okada's Voice Changer, Kaggle
• 🆕 UVR5 UI, by Eddy, ArisDev & Nick088 Kaggle
• 🆕 RVC AI Cover Maker UI, by Shirou & ArisDev Kaggle
• 📖 How to use RVC Mainline on Kaggle by Cauthess

Note: Kaggle limits GPU usage to 30 hours per week.

More commands: -colabs, -hf, -local, -rvc, /help

[amber cairn]

Hey guys! I have this uh interesting little project that if I can't find someone to make for me (commissioned) at least pointed in the right direction in terms of what software or sites to use for it. Nothing sketch just the "template" is a little weird

[elder willow]

I am interested

Is it nsfw?

[ionic pumice]

oh dear

[elder willow]

Real

[amber cairn]

not at all

i mean actually

idk ill let you judge i'ma dm you an insta reel

[elder willow]

Real, i don't have instagram 🔥

Please send the video instead

[river verge]

DIlly ding, dilly dong! A new RegalHyperus drum model just released!
Beat from Melty Blood (Remastered) (Drum model no. 578)

[torn wolf]

hi

i need help for using an ai model

[covert lake]

tell:

• ur pc gpu & CPU
• what u want to do (realtime for calls or ai covers)

[torn wolf]

ai covers

-i7 gen 13

[covert lake]

what about gpu?

[torn wolf]

yes

[covert lake]

what you said is the CPU, I asked also for the GPU

[torn wolf]

rtx 4060ti

[covert lake]

that's great

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

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

If u want, there's also cloud ways that are easier to use, but are time limited:

• Weights.gg: Easiest Possible Ever Automatic
• Ilaria RVC Zero: Fastest free on cloud
• Applio (ui)

I'd suggest locally since you got a good pc

for any issues, use #✨│ai-help

[glass ravine]

[plain cove]

@signal cairn!

[hardy parrot]

Hello people. I make covers with the voice of Conor Mason from NBT. I want to share them with you and hear your opinions and suggestions. How do I do this?

[uneven hawk]

How do I do voice covers. Like i get an scene from a movie or series and make it so the Ai Voice Model says everything the Person in the Scene says?

[light sierra]

luck doesn`exist all good things comes from Allah

[dusky grove]

Do any of you guys have any websites where you can design your tts voice by ai (not elevenlabs.io)

[stiff drum]

working on setting up an AI using ollama through webui. does anyone know how i would access its modelfile.md

ive been skimming through the github page but cant seem to get it

wait.... do i just open a text file in vscode and edit through there and change its file type????

HAHA i think i found it

[solar torrent]

LLM model named Obama.

[elder willow]

for anyone who uses codename rvc fork: does it support refine gan and 441k yet?

[pallid socket]

I need help with applio for ai covers for whatever I can use to make it sound better

[timid mango]

-hf

[red frost]

🤗 Huggingface Spaces

• UVR5 UI, by Eddy and Ilaria Huggingface Spaces
• Ilaria RVC Zero, by thestingerx Huggingface Spaces
• RVC⚡ZERO, by r3gm Huggingface Spaces
• Applio, by IA Hispano Huggingface Spaces

More commands: -audio, - colabs, -kaggle, -local, -rvc, /help

[pallid socket]

Hi

[red frost]

THE FORTNITE BALLS LEADERBANAN

1. Yui — 1
2. BalbalEko(Your Moms Weights) — 1

Feb 8, 2025, 10:15 AM

[bold rampart]

Does this count as cheating, or not?

It's all automatic. I did not move my mouse or interact with anything. The app does it all for me. This is just one of the examples. There are endless possibilities with this!

Until you run out of quota, of course. This uses a LLM model, so yeah, you need an API key, which of course I do have

[hollow tangle]

that became old already

also, furry