Skip to Content


Mediamined is a project I've been working on together with the great folks at Imagine Research for about a year now. With some help from the U.S. National Science Foundation, we're now making public some of our technology.

Mediamined offers the ability to classify sound files into a taxonomy of text labels by analysing the audio signal. We use machine learning techniques to learn the acoustic signatures that correspond to what a bass guitar, male vocals or door slam sounds like, and then classify new sounds according to those signatures. Another capability is to use those signatures to determine acoustic similarity between audio files. This can be used for finding music which is most similar to a query sound file, or to find alternative sound effects, as our CEO Jay explains in a video interview.

Some other write ups have appeared including a nice article on Lockergnome.

How is Mediamined deployed

Dear Leigh,

Just noticed you've been at IRCAM! :-)

I'm quite keen to see a demo of Mediamined but haven't find anything except few videos. How is Mediamined deployed? Is is a software, a client (no local algorithm), or simply a service that requires access to the client database?

I could have a daily use of this and I'm really interested.


Mediamined deployment

Mediamined is an array of technologies, consisting of a large C++ library, which is deployed as Unix or Windows command line tools operating behind a typical web front end, such as PHP apps, or for batch processing of large databases. Alternatively Imagine Research can offer it as a web service, using HTTP PUSH of sound files, and retrieving XML metadata files via HTTP. It can also run as client Mac/Windows applications for standalone operation. At the moment, IR is not offering it directly for end users, but please subscribe to the Imagine Research's news feed for forthcoming announcements.

Copyright | blog