I have created this dataset by taking 50K clean and 50K dirty from people speech and the core purpose is to fine-tune seed-vc model.
