Datasets:

---

k-l-lambda

/

imslp-crawling

like 1

i think that's similar to musicxml

also unrelated but the entire dataset is only 28.91 GB? the metadata says it should be over 1TB? could anyone else confirm?

i think that's similar to musicxml

also unrelated but the entire dataset is only 28.91 GB? the metadata says it should be over 1TB? could anyone else confirm?

Yes, only a subset was uploaded. The subset is the part overlapped with google maestro dataset.

thanks for your prompt reply! do you plan to release the full dataset?

thanks for your prompt reply! do you plan to release the full dataset?

It seems there is a 300GB size limit for a normal hf dataset repo. However you can crawl imslp files by yourself.

You can read the data.db like this script,
and get file links for all works.

thanks for the pointers!

in the dataset the retrieved links are relative links. do you have any rules to resolve to the corresponding absolute links? thanks very much!

(edit: sorry i got the links, thanks very much!)

actually, would you mind sharing the entire dataset somewhere? be happy to pay for it. thanks!

actually, would you mind sharing the entire dataset somewhere? be happy to pay for it. thanks!

Do you mind private chat? Send me your wechat or gmail to me

Did you end up sharing the dataset somewhere? Is it publicly available?

Did you end up sharing the dataset somewhere? Is it publicly available?

I just uploaded the db files in huggingface native format. Now you can access all asset file URLs in https://huggingface.co/datasets/k-l-lambda/imslp-crawling/viewer/Work.

Hope that's helpful for you.

Thanks!