---
license: openrail
---
Don't forget to credit me @Smotto if you do use this model!
Youtube: smotto_ai
Tiktok: smotto_ai

RVC v2 model

Data
  - Singing style taken from her albums, lives, and covers. Cut and clean them manually.
  - 77.4 MB of data == 7 min and 3 seconds
  - 48khz 16bit-depth audio files

Processing
  - Kim vocal 1 -> Reverb HQ -> Karaoke 2 (if needed) -> DeEcho -> Denoise

Hyper-parameters:
  - mangio-crepe
  - 16 batch
  - 16 pitch hop-length
  - 300 epochs
  - 3000 steps

Example: https://www.youtube.com/watch?v=lZ2NuDCgLiU