Training With Mixed Dataset/readme.txt

With further training, the output quality dropped significantly and still could not surpass the RMVPE model, so I decided to stop training.