RVC VOICE BANK
Kuroki Kasori AI is a voice model developed using RVC (Retrieval-based Voice Conversion) technology—an advanced voice cloning technique powered by artificial intelligence. This model allows any vocal recording to be transformed into Kasori’s unique and expressive voice, while preserving emotional nuance, dynamic range, and tonal clarity.
Designed for artistic and creative purposes, Kuroki Kasori offers language-specific versions trained to enhance vocal quality and deliver more accurate results by adapting to the linguistic characteristics of each language. This makes it a versatile and powerful tool for music, storytelling, and experimental content creation.
The model is available for download and creative use. Please make sure to read the Terms of Use section before using it.
The Kuroki Kasori voice models were created to expand their usability and provide additional tools for users who wish to incorporate the singer’s voice into their personal projects and creative works. We kindly ask that they be used responsibly and respectfully to ensure appropriate usage. If any misuse is detected, this RVC-based project may be taken down.
The goal is for these models to complement areas not covered by the existing voicebanks available on other platforms where Kasori is already featured.
General
For RVC
Kuroki Kasori AI V2
Kuroki Kasori RVC V2 is an artificially intelligent trained voice that can replicate acapellas with its own voice. It has a solid tonality with a wide vocal range that allows it to reach very high notes with great power. Its foundation is solid, which means it cannot sing in soft or other tonalities.
This voice bank, compared to its RVC V1 version, has corrected the errors of its database, increasing its vocal naturalness and extending its vocal range.
This Kuroki Kasori voicebank was released on 07/08/2023.
Features
-
Model: 40 hKz - Mangio-Crepe
-
Epoch: 100.
-
Database: 3 hours and 37 minutes.
-
The database was built with 2 hours and 37 minutes of singing and 1 hour of speech. It was constructed based on 4 languages (ESP/JP/EN/KR).
Art by @Poicyss
English
Kuroki Kasori Natural ENG
For RVC
Singing voice model created using English recordings, optimized for accurate results in the language and a natural vocal tone true to the singer.
This Kuroki Kasori voicebank was released on 06/05/2025.
Features
-
Model: 40 hKz - rmvpe.
-
Epoch: 350.
-
Database: 30 minutes of data singing in English
Kuroki Kasori Soft ENG
For RVC
Singing voice model created using English recordings, optimized for accurate results in the language and a soft vocal tone true to the singer.
This Kuroki Kasori voicebank was released on 06/05/2025.
Features
-
Model: 40 hKz - rmvpe.
-
Epoch: 350.
-
Database: 30 minutes of data singing in English
Kuroki Kasori Solid ENG
For RVC
Singing voice model created using English recordings, optimized for accurate results in the language and a solid vocal tone true to the singer.
This Kuroki Kasori voicebank was released on 06/05/2025.
Features
-
Model: 40 hKz - rmvpe.
-
Epoch: 350.
-
Database: 30 minutes of data singing in English