Today happens to be the acceptance date for both ISMIR (nope, delayed!) and WASPAA, and we are very excited to see publications using the MSD finally being released.
One of the most pleasing part of creating a dataset is discovering new ways of using it, seeing what people made out of it. To aggregate those ideas, we are starting a new informal web page reporting papers that use the MSD:
http://millionsongdataset.com/pages/publications
Please send us an email with a reference to your work if you think it should be included. We'll eventually parse ISMIR and other main conferences to find such work, but the full program will not be released for a while.
On a bragging note, we are proud to announce our paper on large-scale cover song identification at WASPAA. I'd be surprised if we were the only ones working on this, so if someone else has results on the Second Hand Songs dataset, we'd love to talk! Regarding ISMIR, we did see the temporary grades during the discussion phase, but let's wait until Tuesday before we celebrate.
Related to these two papers, we're updating the github repository, adding the code to reproduce our experiments from ISMIR (year prediction) and WASPAA (cover songs). If you want your code to be part of the main MSD github as well, write us!
A few random notes:
- happy birthday to The Echo Nest, 6 years already!
- there might be a few additions to the MSD announced at the ISMIR tutorial, or slightly before... stay tuned!
- WOMRAD deadline soon
- kitten fighting apples
-TBM
- millionsong's blog
- Login to post comments