WebKit

Introducing the RVC V2 Model, specifically tailored for 's iconic voice as Vinyl Scratch from My Little Pony/Nowacking. This version, trained on audio from fan-made animations and a 2012 song, has been fine-tuned for 300 epochs, resulting in impressive AI music capabilities. Ideal for creating AI covers or even text-to-speech projects, particularly fitting for alternative rock genres. The model employs RMVPE for pitch extraction, offering remarkable authenticity. Give it a try for your next project – visit our website for free AI tools and learn more about us.

< Nowacking's (< voice for Vinyl Scratch from around 2012, sourced from cleaned up audio from fan-made MLP animations and a song he made around the same time pre-transition. Works fairly great for singing, can imagine a use for this in an alternative rock song. Likely going to train this a little further but for 300 epochs it's not too terrible. RMVPE for pitch extraction. If you end up using this in a song or whatever, feel free to tag me on my YouTube at: < Don't feel obligated to, though! As usual, more details are at the Hugging Face repo this is hosted on.

Vinyl Scratch (My Little Pony/Nowacking)

Vinyl Scratch (My Little Pony/Nowacking) (RVC V2) 300 Epochs

jenDEV

Also paging <@283399460036542464> <@268913577711108097> <@227976076624789504> as they've requested this at some point. 👽

Aside from some general mixing and etc, it's directly from RVC

Honestly as long as you have clean audio data to work with, you should be good. Training with a decent epoch count is encouraged too (I find 200-300 is a standard base for most models I've noticed)

It's also optimal if the sources are the same/nearly identical, otherwise the final model ends up "overlapping" with itself sometimes (learned this from experimentations with a Season 13 Hank Hill model I was messing around with.)

I feel like anything below like 1 minute of data usually works better with around 100-150

xhepyxopila

Sounds amazing, you earned my precious Like 🔥🔥🔥

Acorn Sweet

The chef from Octodad! Glad to see the server back, been wanting to share this for a little bit. < RMVPE for pitch extraction, more details on the Hugging Face repo. Credit not required, but appreciated!: <

Chef Fujimoto (Octodad: Dadliest Catch)

Chef Fujimoto (Octodad: Dadliest Catch) (RVC V2) 200 Epochs

"Nobody suspects a thing!" < Been a little while since I've done anything RVC-related, wanted to try something fun. Octodad's blubtastic voice from the 2014 PC game "Octodad: Dadliest Catch". Model trained on a little over 4 minutes of his blubs at 200 epochs, not including any dialogue from the DLC (the mixing was done differently for it). Works best if you imitate his nonsensical blubs and set the search feature ratio to 1. RMVPE for pitch extraction, extra information on the model card for the Hugging Face repo this is hosted in. Credit not required but appreciated, if you decide to do anything with this on YouTube feel free to mention me: < Interested in seeing what can be made of this! Tagging since this was a request at some point. I'll be posting a Chef Fujimoto model from the same game sometime later.

Octodad (Octodad: Dadliest Catch)

Octodad (Octodad: Dadliest Catch) (RVC V2) 200 Epochs

"Something something something, Right? Listen!" < Model trained on around 10 mins of one of GradeAUnderA's drama videos from 2016. Out of a range of epochs, 300 sounded the best. I think a model trained on his 2013 videos as well a model trained on his newer content would be something to explore in the future... for now, here's 2016 Grade. Harvest for pitch extraction, more details on the RVC repo that this model is in. I saw someone earlier submit a Leafy model... 2016 is making a come back, isn't it? Also paging as it seems they requested a GradeAUnderA voice model a little while back.. so enjoy I guess lol.

GradeAUnderA (2016 era)

GradeAUnderA (2016 era) (YouTuber) (RVC V2) 300 Epochs

(Updated with a Hugging Face link, will be using that from now on.) Drive: Hugging Face: The classic Worms speechbank, trained on a dataset directly extracted from Worms 3D. Extremely minimal, but interesting how it turned out! Definitely doesn't appreciate lower notes.. I'd say a good pitch range if you're using lower voice inputs is somewhere around 6 - 12, as this is a pretty high pitched voice. Trained by myself, if you'd like to credit me just link/tag my public YouTube: < Harvest for pitch extraction. (Language warning on the recording test)

Vinyl Scratch (My Little Pony/Nowacking)

Description

Samples

Saiba Momoi (Blue Archive)

Adel imam (Egyptian Actor) عادل امام (ممثل مصري)

JP Furina / Focalors CV: Inori Minase (Genshin Impact)

Ariana Grande AI

Hu Tao JP (Genshin Impact)

JENNIE of BLACKPINK [Strong Ver.]

Saiba Momoi (Blue Archive) (VA: Tokui Sora)

Hatsune Miku

SpongeBob SquarePants (Talking And Singing)

Takanashi Hoshino (from Blue Archive)

Satoru Gojo (JJK) [VA Yuichi Nakamura]

ENHYPEN Heeseung

Sunaokami Shiroko (Blue Archive)

Goku (Mario Castañeda) (Latin American Spanish) (UPDATE)

Villager (Minecraft)

Mortis [Brawl stars]

Jungkook (BTS)

Melodie - (Brawl Stars) - (English & Korean Voices)

Tendou Arisu (Blue Archive)

Kanye West