custom training data split

[juheo]

hello. Thanks for sharing your great research!

I'm going to do an experiment to see the effect of data size on diffsinger model training.

So, I want to train using only a part of the opencpop data, but I am having a hard time figuring out how to use only the specified segment file for training.

Can I just make the desired data exist in data/raw and then run binarize.py?

It would be a great help if you could give me an answer.

thank you!

[hsk-yjk]

Has this problem been solved? If yes, how did you solve it? thank you

[lampv33]

Have you done the experiment yet? How does the data size affect the quality of Diffsinger model (midi version)?