Fingerprinting Music

MusicIP, based in California, US, announced last week that it had received a US patent for its method of automatically identifying, or “fingerprinting”, digital music files.

MusicIP, the first global search engine for music, has been awarded U.S. Patent #7,013,301 for its Audio Fingerprinting System and Method that provides reliable, fast, and robust identification of digital musical tracks.

The technology in the patent identifies digital music tracks at the master-recording level, by the actual sounds in the track. Released as MusicIP’s affordable Open Fingerprint™ service, the method identifies the same piece of music, consistently, anywhere in the world regardless of language or format.

With the Open Fingerprint Architecture, digital tracks can be identified consistently against MusicIP’s Music Digital Naming Service (MusicDNS) dataset of more than 17 million analyzed songs, providing a simple, dependable Web service to identify music tracks and provide basic metadata. Under the Open Source license options, the public-domain track metadata returned by MusicDNS can be used freely for any application needs.

How It Works :

The system can recognise a song from its audio “fingerprint” in a fraction of a second. This allows users to rapidly organise their music collection, discover more about a particular track or get new recommendations, through connected databases, regardless of the format of the audio file.
Dominant tones

To make a fingerprint, MusicIP quickly scans the first 2 minutes of a track and records frequency data every 185 milliseconds, before compressing the results into a 512 byte file. It also measures records the four most dominant tones in the first 30 seconds of the music.

The program uses information about these dominant tones to narrow the search before searching the song database using the frequency information. Dunn says this allows the company to perform hundreds of searches each second and that the service is sensitive enough to distinguish between different versions of the same tune, such as live and studio recordings.

Rivals :

While other companies use digital fingerprints to identify songs, the databases they claim are much smaller. Shazam Entertainment, in London, UK, runs a service that lets users hold their mobile phone up to a music source for 30 seconds, and receive a text message identifying the song and artist. But Shazam has a database of only about 2 million songs.

Gracenote, in California, US, is a leading provider of online music identification but claims to have a database of around 7 million songs.

An alternative approach is user collaboration. Online services such as Audioscrobbler and Pandora, for example, recognise songs and make recommendations by searching through user-generated playlists.

But Dunn hopes the speed and size of MusicIP’s database will make it stand out. Companies must pay a license fee to access MusicIP’s Music Digital Naming Service service, but non-profit organisations can access it for free.