first 32k model, and i think it sounds alright. trained on the same dataset as the 40k one steps: 18k huggingface: https://huggingface.co/sxndypz/rvc-v2-models/resolve/main/shiina_32k.zip

samples

and kit if you're planning to update the space by adding shiina, just use this version it's better

Amanogawa Shiina (Phase Connect / VTuber) (RVC v2 32k) - 420 Epochs

a

next is memory (yes, this is the actual name) from phase connect hoorayyyyyyy