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