Million song dataset available for download | Hacker News Reader

Million song dataset available for download

flowingdata.com

48 pointsnreece11 comments

Direct link: http://labrosa.ee.columbia.edu/millionsong/

If you need an API as an alternative http://developer.echonest.com/docs/v4/

That's the EchoNest API and that's where the database was collected from.

This is very interesting data and just looking at it has sent me into a ponder-mode as to what to do with it.

Also, an interesting note on the EchoNest site about the API usage: The Echo Nest's APIs are for non-commercial purposes only. If you are a lone developer selling advertising that is barely covering your running costs, we consider that non-commercial

Once you click through, you find out the dataset is.. 280GB. So you might want to steer clear unless you have the ability to deal with that ;-) There's a 1.8GB subset also.

Or, if you're just interested in the metadata, that dataset is available too. See the FAQ http://labrosa.ee.columbia.edu/millionsong/faq "What are song / aggregate / summary files?" question. (It's a 300MB file)

This thing is dying to be distributed via torrent.

Not really. Interest for that file is very low so few seeders if any at all (think 12 months from now). Few people have the bandwidth to transmit so much data quickly.

This is better off on a good http host.

BT doesn't have to provide for the long-term, in the short term it can help with not making the user download a bunch of small files.

We planned to release a R wrapper and looked at the default HDF5 library for R on Ubuntu. Unfortunately, it crashes on empty arrays. This happens when a track has no musicbrainz tag, for instance. If any R specialist is willing to help us with this, please contact us!

I'm sure someone from HN could help with that...

The site is down.

Can someone please post more about it? What's the schema for the data? How clean is it? What are they trying to achieve with this release?

Thanks!

I dug around for some cached files and found the following:

Why: http://webcache.googleusercontent.com/search?q=cache:H52osT6...

Getting it: http://webcache.googleusercontent.com/search?q=cache:0I8CMo0...

from the second link I found it's hosted here:

http://www.infochimps.com/collections/million-songs

where you can download a subset as well as the whole thing. I'm downloading the subset now, so I can't comment on the cleanliness or schema yet.

Site is down. If someone manages to get the metadata file and seeds it, please post a link here!

If we assume the average song is 3 minutes, that's nearly 6 years of continuous listening...

Reply on news.ycombinator.com