Kanye West - RETRAINED
RVC v2500 epochsArtistEnglish

๐Ÿ“ฃ

From the creator

"I have a patreon where you can support my RVC stuff (colabs guides models etc) and also get access to direct tech support on how to use the models: Link to the 40khz version (I think I prefer this): Link to the 48khz version: Retrained with a better crisper dataset courtesy of and trimmed down by hand cleaned up and further polished by me to the best 20 minutes of data that there is. I sent the dataset to nano so it could be trained by him on his fast AF GPU. Enjoy!"

Description

Introducing the Kanye West RVC v2 (Crepe) - a retrained AI voice model utilizing Retrieval-Based Voice Conversion (RVC) technology. This sophisticated model has undergone 500 epochs of training, resulting in enhanced accuracy and clarity when converting text input into speech. With striking similarities to the iconic rapper's unique vocal style, this RVC model excels at creating authentic and engaging AI covers. As a leader in artificial intelligence innovation, Weights provides cutting-edge solutions like the Kanye West RVC v2 (Crepe), empowering creators to generate high-quality AI music covers effortlessly. Additionally, our advanced text-to-speech capabilities enable businesses and individuals alike to harness the power of AI voices for various applications. Join our community on Patreon to stay updated on the latest developments, collaborations, tutorials, and more within the realm of RVC technology. Unlock exclusive benefits such as priority access to new models and direct support from our knowledgeable team. To start exploring the possibilities offered by our state-of-the-art AI tools, sign up for free today and begin creating your own AI

See more

Weekly Metrics

Views

32

Creations

23

Downloads

20

Audio Samples

๐ŸŒŽ All

#1

Male

Speaking

๐Ÿ‡บ๐Ÿ‡ธ English

#2

Male

Singing

๐Ÿ‡บ๐Ÿ‡ธ English

#3

Male

Singing

๐Ÿ‡บ๐Ÿ‡ธ English

#4

Male

Singing

๐Ÿ‡บ๐Ÿ‡ธ English

#5

Male

Singing

๐Ÿ‡บ๐Ÿ‡ธ English

#6

Male

Singing

๐Ÿ‡บ๐Ÿ‡ธ English

#7

Male

Rapping

๐Ÿ‡บ๐Ÿ‡ธ English

#8

Female

Speaking

๐Ÿ‡บ๐Ÿ‡ธ English

#9

Female

Speaking

๐Ÿ‡บ๐Ÿ‡ธ English

#10

Female

Speaking

๐Ÿ‡บ๐Ÿ‡ธ English

#11

Female

Speaking

๐Ÿ‡บ๐Ÿ‡ธ English

#12

Male

Speaking

๐Ÿ‡ช๐Ÿ‡ธ Spanish

#13

Male

Speaking

๐Ÿ‡ซ๐Ÿ‡ท French

#14

Male

Speaking

๐Ÿ‡ฎ๐Ÿ‡น Italian

#15

Male

Speaking

๐Ÿ‡ฐ๐Ÿ‡ท Korean

#16

Male

Speaking

๐Ÿ‡ท๐Ÿ‡บ Russian

#17

Female

Speaking

๐Ÿ‡ช๐Ÿ‡ธ Spanish

#18

Female

Speaking

๐Ÿ‡ซ๐Ÿ‡ท French

#19

Female

Speaking

๐Ÿ‡ฎ๐Ÿ‡น Italian

#20

Female

Speaking

๐Ÿ‡ฐ๐Ÿ‡ท Korean

#21

Female

Speaking

๐Ÿ‡ท๐Ÿ‡บ Russian

User Creations

Not certified by Weights

"์ ˆ๋Œ€๋กœ ์ง€๊ธˆ๊ณผ๋Š” ๋‹ค๋ฅธ ์ž์‹ ์ด ๋˜๊ฒ ๋‹ค๋Š” ์ƒ๊ฐ์„ ํ•ด์„  ์•ˆ ๋ผ. ๋„Œ ์นด๋‚˜๋ฉ” ๋งˆ๋„์นด, ๊ทธ๋Œ€๋กœ๋ฉด ๋ผ. ์ง€๊ธˆ๊นŒ์ง€์ฒ˜๋Ÿผ, ์•ž์œผ๋กœ๋„."
"Good evening. Today, our fellow citizens, our way of life, our very freedom came under attack in a series of deliberate and deadly terrorist acts. The victims were in airplanes, or in their offices; secretaries, businessmen and women, military and federal workers; moms and dads, friends and neighbors. Thousands of lives were suddenly ended by evil, despicable acts of terror."

Model Files

Compressed File Verified

Kanye West - RETRAINED.zip

md5: cc2d2fd6bdd2aa785f04a4d05125d8e9

Individual Files Verified

model.index

md5: d610ecb59580ab9ae9fe941fc1464f35

195.42 MB

model.pth

md5: 58e7229f6562ef63e45bc032a1dfe0cb

52.67 MB

Comments

Kalo whereโ€™s the sample

This is all lossless quality stems used for training mainly the earlier eras so more rap but some singing. It's still better at singing than the other model.

Audio removed in response to copyright claim.

Audio removed in response to copyright claim.

i hope you die

thank you โค๏ธ

there goes mine lmao

mine might be better in some singing scenarios

this is primarily early era rap it was trained off

but it's still good at singing for sure

48khz for comparison. All of these are on harvest. Reducing reverb with UVR and denoising as well as using mangio crepe at low size would likely be a more fair comparison. Maybe 40khz is better sometimes? Dunno

Audio removed in response to copyright claim.

Another comparison of 40khz vs 48khz:

Audio removed in response to copyright claim.

Audio removed in response to copyright claim.

Audio removed in response to copyright claim.

Audio removed in response to copyright claim.

As a music producer I was surprised that 40hz is the default. While Iโ€™m writing this my first 48khz model is training!

We just got 48k training so that's why 40k is the default.

Oh didnโ€™t know that

Makes it even worse that RVC v3 is in development just as we get RVC v2 48k.

As long as we have RVC im happy ๐Ÿ˜‚

Because 48khz being better is kind of debatable in this context. and sometimes is straight up worse. having more data than is necessary for training can make it sound more rough. 100%

Honestly 48k has been very good in my experience.

If a model is iffy its usually due to Mango Crepe or how iffy the dataset is.

48khz was better for my 2 minute studio quality kanye dataset and worse for my 20 minute studio quality dataset. all of it was of consistent quality

it depends 100%.

did you resample your dataset in 40k if you want to train it on 40k model?

i did it on my own model and it sounds flawless and clean

in which procedure? preprocessing? feature extraction? my previous models sound robotic when my dataset was 44.1k sp and trained on 40k

i tried 48k and i think it's experimental it sounds robotic on high pitched notes using mangio-crepe method

maybe pre-trained model that i used on v2

it hasn't but can handle very well on 40k model that trained with 40k sr dataset

i will look at it on my next training

yeah the more dataset you have fed in the more features it learned from it

my 10-minute dataset can also make my model sound in a reasonable way as long as it's clean and less noisy

i hope that too

yeah i have my 48k dataset ready for that just in case ๐Ÿ˜…

Why would you explain that here

Go explain it in <#1089301011149103276> or dms or somethin next time

sowwy ๐Ÿฅบ๐Ÿฅบ

it does that for you in RVC's preprocessor

oh i didn't know that. thanks for info

Video removed in response to copyright claim.

thats realtime btw^

As in you spoke the lyrics?

i was bored

kanye singing 20 min by lil uzi vert

Audio removed in response to copyright claim.

HOLY SHIT THAT'S ACTUALLY SO GOOD PLEASE DO ANOTHER UZI SONG WITH KANYE

some songs should not be touched

whar do you mean

this is a joke

Audio removed in response to copyright claim.

LMFAOOOOOOOOO

happy 50 fire emojis

Audio removed in response to copyright claim.

Add a comment

Post
Selected Audio
Selected Audio