Update to Echoprint database hosted at The Echo Nest
We have updated the Echoprint database with a fresh set of over one million popular songs for search using Echo Nest song/identify API: http://developer.echonest.com/docs/v4/song.html#identify Be sure to specify “&version=4.1x” on your queries to indicate the Echoprint codegen version (currently 4.12).
We provide a human-readable list of all songs that are in this database for testing and evaluation purposes:
http://echoprint-data.s3.amazonaws.com/list_of_songs.txt [Warning! Size is 109 MB]
Each line in the list has the following format:
Echo Nest song ID —- artist name —- song title
The Echo Nest song ID links each song in to The Echo Nest API and is useful for evaluation purposes.
4:21 pm |
April 22 2013
Human-readable list of tracks
I have uploaded a list of tracks contained in the Echoprint database to our server and you can check it out via http://echoprint.me/data. (This is much easier for humans to read than the raw JSON dumps, and it is only 11 MB.)
12:04 pm |
March 19 2012
Echoprint Codegen 4.12 released
The latest release of Echoprint Codegen, version 4.12, is now available for download.
This is a stable maintenance release consisting of bug fixes and documentation updates. There are no changes to the fingerprint codes which are output by the codegen.
It is available from GitHub via the “release-4.12” branch at http://github.com/echonest/echoprint-codegen/tree/release-4.12
The following command line invocation will download Echoprint Codegen using Git:
$ git clone -b release-4.12 email@example.com:echonest/echoprint-codegen.git
This method clones the repository and then switches HEAD to the “release-4.12” branch. The software can then be compiled and installed from there according to the instructions in README.md .
Alternatively, and perhaps more easily, you can download a self-contained zipball or tarball from http://github.com/echonest/echoprint-codegen/tags The tag is “v4.12”. Downloading the software this way does not require Git.
1:14 pm |
March 15 2012
Welcome to Echoprint! / FAQ
We’re very excited to announce Echoprint to the world. Echoprint is in its infancy. We’re ready to release it to the world for feedback and general use.
What does this do?
Echoprint is a music fingerprint or music identification service. It listens to music signals and tells you what song is playing. It’s backed by a huge database of music that grows with the community and further partnerships. On launch we’ve partnered with Musicbrainz.
Who is this for?
Echoprint is a suite of tools for developers and the music industry. There is no “Echoprint app” in the app store, although we’re sure within weeks there will be. This is for developers of music applications. If you’re building a mobile music experience, you’ll want to use this. Or if you’re sitting on top of a large database of music. Or if you want to deduplicate your collection, do copyright detection, or resolve your users’ catalogs to yours.
Really open source? Can I use this commercially?
Yes and yes. The code generator is MIT and the server is Apache 2. There is absolutely no restrictions to using the code generator or server in your app. Check them out on GitHub. If you use our data, you’ll need to read the data license — but it only says that if you collect new fingerprints, you have to contribute them back to the community.
Does it work “over the air”, identifying songs over a microphone?
Yes - Echoprint has been designed from the ground up for OTA, and our informal tests have demonstrated many successful and promising results for this scenario. The system still needs a little more tuning, however, and is under continued development to further improve accuracy and performance.
Can it scan a file to get the correct metadata?
Yes! From anywhere in the file, at least 20 seconds of audio signal is needed.
Is this ready to go? Is it mature? How can I trust you guys?
It’s ready to go, but check our status page for some important notices. One main caveat: OTA eval is not yet finished, although it is promising. We’ve been running fingerprinting at serious scale for developers and our customers for over a year now. We know about all the problems and scale issues. The Echo Nest is in the business of making everyone’s lives more awesome.
Do I have to boot my own server?
No. We expect most people to use a 3rd party hosted lookup server. The Echo Nest hosts one under the song/identify API. For example, the following song/identify call:
resolves to the right song ID:
For an example of how to construct song/identify calls using codegen, read here.
Does it scale?
Yes. Echoprint’s closed source brother, ENMFP, has been in wide use for about two years and has close to 60 million tracks on a single server. Echoprint uses the same back end. Depending on your architecture, a single box can match 50 queries a second. The code generator is blisteringly fast— you can scan enough audio for a query in less than a tenth of a second.
What do I do now?
Join the discussion group. Download the codegen or server from GitHub and build it.
What’s the deal with Musicbrainz?
Musicbrainz will be working on integrating Echoprint into their world for track resolution. They have a test server ready to go that understands Echoprint codes.
This is awesome / I want to help!
Yes! Check out the code and file pull requests or issues. Get a job at the Echo Nest (we are hiring like crazy.)
9:57 pm |
February 7 2012
| 1 note
We will be updating this page regularly with status updates.
We currently provide data from a small collection of popular music in our data dumps, roughly 150,000 songs. The syncing mechanism for outside catalog is under development.
QA / Evaluation Status
See our “bigeval” metrics on How it works. Echoprint is “almost perfect” on file matches and de-deuplication. We have not yet fully evaluated OTA performance other than a few dozen in-the-field tests. Please file an issue on GitHub in the server project for any match issues. All further QA will stay at the server match level.
Codegen 4.12 is “gold” — no further algorithm changes will be baked in until a major revision. We welcome pull requests and issues on codegen for compilation and speed optimizations only.
Echoprint-server master is booted at the Echo Nest and serving queries for song/identify. We will be making changes to best_match_for_query for OTA fixes as above but otherwise the server is ready to go.
9:55 pm |
February 7 2012
| 2 notes