Mr. Krabs (SpongeBob)
RVCFictionalEnglish

πŸ“£

From the creator

"This model was trained on the few Clancy Brown video game voice lines I could find. In total it is roughly around 10 minutes or so of audio. It was trained using the crepe method on RVC. Should work well with vocal impressions and possibly singing but has yet to be tested. Hugging Face Mirror:"

Description

Introducing the innovative Mr. Krabs (SpongeBob) RVC 1.25K Epochs [56.25K Steps] - an advanced AI Voice Model powered by Retrieval-Based Voice Conversion (RVC) technology. Exclusively developed at Weights, this unique model captures the distinctive voice of renowned actor Clancy Brown, famously known as Mr. Krabs from the beloved animated series SpongeBob SquarePants. By utilizing our cutting-edge RVC technique, we've meticulously trained this model through 1.25k epochs, resulting in astonishingly authentic audio outputs. The Mr. Krabs RVC model offers incredible versatility; generate realistic AI music, produce engaging AI covers, or transform text into lifelike speech using our sophisticated Text-to-Speech functionality. Leveraging over 56.25k steps, our skilled team has fine-tuned this model to perfection, ensuring unparalleled precision when replicating Mr. Krabs' iconic tone. Unlock limitless creative possibilities – use our state-of-the-art AI tools to bring

See more

Weekly Metrics

Samples

New
Classic
TTS

Pitch

0

1

Male

Singing

πŸ‡ΊπŸ‡Έ English

2

Female

Singing

πŸ‡ΊπŸ‡Έ English

3

Female

Singing (Dry)

πŸ‡ΊπŸ‡Έ English

4

Female

Singing (High)

πŸ‡ΊπŸ‡Έ English

5

Male

Singing 2

πŸ‡ΊπŸ‡Έ English

6

Male

Singing (Dry)

πŸ‡ΊπŸ‡Έ English

7

Male

Singing (Dry, High)

πŸ‡ΊπŸ‡Έ English

Comments

Uncut Sample Compilation of SpongeBob Clips (audio was not used in dataset)

Audio removed in response to copyright claim.

Original Sample made by me.

Audio removed in response to copyright claim.

Mr. Krabs singing "Money" By Pink Floyd

*The sax section bugs out but its kind of funny. He turns into an instrument lmao.

I put the new forgis on the jeep! (Ballin' Rap Test)

what settings do you use for the generations? my samples sound innacurate

my childhood is ruined

how did you train for more than 10000 Epochs?

One memory at a time. I'm probably gonna train all of the main cast.

Open infer-web.py from the rvc directory in notepad or a code editor and ctrl+f search for 1000 and edit it to whatever you want the max amount the slider to go to.

Wowww.. Thanks for that. That 1000 imit was stupid. So is limiting the batch size to only 40 when the GPU has ample memory

Lastly. Where did you get the "crepe" model from? It isn't in the RVC beta 2023-05-13 zip file

Mangio RVC is a fork that supports it it is experimental but works fine most of the time I fixed the code and uploaded it for convenience :

this fork also has unlimited epochs by default

also you can track where the model begins to 'overtrain' in tensorboard

Mr. Krabs Sings The Campfire Song Song

Audio removed in response to copyright claim.

someone do a Mr. Krabs cover of Cartier Gucci Scarf by Young Thug since he basically sounds like Mr. Krabs at one point on that song anyway

Mr. Krabs cover of Cartier Gucci Scarf by Young Thug (Damn he popped off fr)

Audio removed in response to copyright claim.

here with me - d4vd DO THATTT

the crazy part is that the section I'm talking about doesn't even sound that different. Young Thug got a mean Mr. Krabs impression

Well not necessarily. In 99% cases training will start to overfit under 500 epochs meaning you shouldn't keep training because it will just get worse. The batch size limit also protects the model. You'll get training speed boosts with the higher batch but the accuracy starts to suffer the higher it is.

Better late than never.

Audio removed in response to copyright claim.

DEJA VU! (Initial D)

Audio removed in response to copyright claim.

LMAOO

who made bro sad smhhh

Imagine putting the voice clips from battle for bikini bottom true the ai

HOLY SHIT GUYS! Thank for Over 100 Reactions! Due to the success of this model there is a high chance it will be retrained at some point on RVC v2! Posting a new random sample of the KRUSTY KREW ANTHEM (BACK ON THE GRILL) by Kash Krabs using my AI model for now.

Audio removed in response to copyright claim.

This would be nice I've noticed this model tends to....capture certain things as if it were a little overtrained.

Sorry yall been working on commissions for a while so I haven't been very active on here I will extend the dataset of this model with mr krabs songs from the show and show audio to help with the dataset length and will train it again. Hopefully before the end of July or whenever RVC v3 is out whichever comes first.

I can't believe this model's been up for almost 2 months and no one has thought to convert fake krab's voicelines from Battle for Bikini Bottom into AI krabs. Good thing I was here for that

Add a comment

Post

Privacy

β€’

Terms

β€’

DMCA