question about parse_data.py
Akiiiira opened this issue · comments
Hello. I'm interested in your great study, and I have 3 question about to create musescore dataset.
According to the preprocess.py in musicnet, at first we loading the musicnet dataset by
[data = np.load('./data/musicnet.npz', encoding='bytes').]
1st: About MuseScore, what is the variable corresponding to the data?
2nd: what will we do by parse_data.py? this create the dataset about MuseScore?
3rd:when I run download_fule() after running parse_url() and search_score_oj(), [BeautifulSoup(xmloj).find('secret')] become NoneType object so [.text] can't run correctly.
why this error occur ?
Originally posted by @Akiiiira in #5 (comment)
Hello Akiiiira,
Thank you so much for your interest. The parse_data.py originally contains the code to only download data through Musescore API. However, it seems Musescore no longer provides the free public API. Although 'parse_url()' still works as a parser, other two functions which use their API to download data can't work anymore. If you need a good quality synthesized dataset, I suggest looking into Slakh (https://github.com/ethman/Slakh). I am also working on this dataset, might push the code later.
Thank you for your reply !
I'm so sorry that I can't download Musescore dataset, but I'll try to look for another dataset.
I'm also looking forward to your new description!