WebKit

Introducing the Nanahira RVCv2 Model, designed for creating AI covers and AI music! With its powerful 48khz sampling rate and extensive 250 epoch training, this model stands out from the crowd. Experience the richness of sound as it brings to life basslines and more, inspired by diverse content spanning multiple eras. Give your creative projects a unique touch with AI-generated music powered by the Nanahira RVCv2 Model.

Bass (line line line). This is Nanahira. Trained on various content from various eras, about 13 minutes of data.

Nanahira

Nanahira | RVCv2 48khz 250 epochs rmvpe

soulnull

about 10 minutes of training data from various show clips < <

Art Bell (radio host)

Art Bell (radio host) RVCv2, 150 epochs 48khz

Came out pretty decent considering she really doesn't say a whole lot.. down with the mafia! < / <

Hat Kid

Hat Kid (RVC v2) 400 epochs 48khz

Brak from The Brak Show/Cartoon Planet. < / <

Brak 200 2pachs

Brak (RVC v2) 200 2pachs

Hello everyone, this is.. running on empty...... food review. training/dataset by soulnull. New version. Samples of the new model further down. < / <

ReviewBrah / The Report Of The Week v3

ReviewBrah / The Report Of The Week v3 (RVC v2, mangio-crepe) 250 epochs 48khz

~~55 minute~~ ~~more carefully curated 26 minute~~ 17 very carefully curated dataset+training by soulnull for all intents and purposes, the new version is the one you want. the old versions still included just in case you want them for whatever reason. samples are down in the thread, the ones here are for 1.0. < / < Trained to 600 epochs with harvest (mangia-crepe really doesn't like his voice), 48khz, rvcv2, all theblahblah stuff.. samples to be posted further below. Old versions: 2.0, slightly better speech compared to 1.0 but some accuracy issues: < / < 1.0, gritty/grainy but soulful music. Not great sounding but fun to play with. < / <

Nanahira

Description

Samples

Nanahira