Name: Keshi | AI RVC Model
Brand: Weights
Rating: 4.4 (20 reviews)

WebKit

 Introducing the Keshi (RVC V2) (300 Epoch) - the latest advancement in AI Voice Models from Weights. This state-of-the-art model has been isolated and trained using all songs from Keshi's album 'always', allowing it to accurately replicate his unique vocal style through Retrieval-Based Voice Conversion (RVC) technology. With 300 epochs of training, this model delivers highly accurate and natural sounding results, making it perfect for creating AI covers and text-to-speech applications.

Our team at Weights has harnessed the power of AI music to bring you an innovative solution that pushes the boundaries of voice conversion models. By using cutting-edge techniques such as RVC, we have created a powerful tool that allows musicians, producers, and developers to explore new possibilities in audio production.

Whether you're looking to create realistic AI covers of your favorite artist or add lifelike speech capabilities to your application, our Keshi (RVC V2) (300 Epoch) model is here to help. And best of all, you can get started today for free! Simply sign up

Isolated and trained by me using all songs from his album 'always'. I was able to isolate his vocals very well using multiple alogorithms with the FLAC versions of his songs. I did my best to remove all background vocals and reverb so it should sound pretty dry. Please credit me if used publicly anywhere such as social media etc. Discord: BigSoulja#8888 TikTok: @souljavr YouTube: You can credit any one of the above handles, does not need to be all of them. Model spec: 300 Epoch, 16 Hop Length, 40K depth, mono input, 14 minute dataset without silences Model

Keshi

Keshi (RVC V2) (300 Epoch)

bigsoulja

Gonna leave it training to higher epochs but it seems to be overtraining past 300 atm

First Keshi model to exist lets go <:LFG:1093368528637276191>

Chase Atlantic - Into It (KeshiAI) https://vm.tiktok.com/ZGJxh2nap/

Amzing work! can you please do Ruel? i tried but failed T_T

vampy

Please credit me if used publicly anywhere such as social media etc. Discord: BigSoulja#8888 TikTok: @souljavr YouTube: You can credit any one of the above handles, does not need to be all of them. Model info: RVC V2, 40K Sample Rate (will redo with higher sample rate when RVC updates to support this), 2000 Epoch, trained with CREPE 16 Hop Length

Pop Smoke

Pop Smoke (RVC V2) (2K EPOCHS)

Based on leaked pro tools sessions, vocals sound dry. Please credit me if used publicly anywhere such as social media etc. Discord: BigSoulja#8888 TikTok: @souljavr YouTube: You can credit any one of the above handles, does not need to be all of them. Model info: RVC V2, 40K Sample Rate (will redo with higher sample rate when RVC updates to support this), 300 Epoch, trained with CREPE 16 Hop Length

Machine Gun Kelly | MGK

Machine Gun Kelly | MGK | (RVC V2 300 Epoch)

Based on leaked pro tools sessions, vocals sound dry. Please credit me if used publicly anywhere such as social media etc. Discord: BigSoulja#8888 TikTok: @souljavr YouTube: You can credit any one of the above handles, does not need to be all of them. Model info: RVC V2, 40K Sample Rate (will redo with higher sample rate when RVC updates to support this), 400 Epoch, trained with CREPE 16 Hop Length

Fivio Foreign

Fivio Foreign (RVC V2 400 Epoch)

Created by me. Based on leaked pro tools sessions, vocals sound dry. Please credit me if used publicly anywhere such as social media etc. Discord: BigSoulja#8888 TikTok: @souljavr YouTube: You can credit any one of the above handles, does not need to be all of them. Model info: RVC V2, 40K Sample Rate (will redo with higher sample rate when RVC updates to support this), 1900 Epoch, trained with CREPE 16 Hop Length

Young Thug

Young Thug (RVC V2 1.9K Epoch)

Created by me. Based on multiple pro tools studio sessions, without silences it is over an hour of pure peep vocals. This is by far the cleanest and most accurate model of him. The vocals sound completely dry and clear. I also used the room tone/noise samples from each pro tools session to perfectly isolate the vocals and remove any slight background hiss/headphone leakage etc. It took me very long to create this dataset with the sheer amout of vocals I had to check through but I think it was worth it. Please credit me if used publicly anywhere such as social media etc. Discord: BigSoulja#8888 TikTok: @souljavr YouTube: You can credit any one of the above handles, does not need to be all of them. Model info: RVC V2, 40K Sample Rate (will redo with higher sample rate when RVC updates to support this), 500 Epoch, trained with CREPE 16 Hop Length those who were gatekeeping peep models, you suck!

Keshi

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