I just finished training Anya RVC model, I used her recent zatsu for the dataset supplemented with clean parts of her cover songs. The model struggles with high notes due to the lack of high notes in the dataset, I need clips or timestamps of Anya singing in high notes to further train the mode.
You can try it directly here, but I recommended running it on google colab to avoid space's free tier limitation
https://huggingface.co/spaces/Wafer22/rvc-modelsDemo/my first attempt at making AI cover
https://youtu.be/KMi1EdeYlHc