Follow Slashdot blog updates by subscribing to our blog RSS feed

 



Forgot your password?
typodupeerror

Comment Re:One Way Audio Fingerprinting Works (Score 1) 141

George Tzanatakis has an open source framework for audio analysis. Amongst the things his framework does is accurate automatic genre detection, mfcc, fft and beat detection.

He also developed the fingerprinting for moodlogic which works.

Philips also have a paper describing a system for fingerprinting, they can detect a song based on any 5 second sample.

I think it is fairly urgent that an open source fingerprinter is released (the music brainz one is by relatable and is closed and fairly inncacurate cause it only uses one feature vector instead of multiples)

Accurate fingerprinting will be useful for:

1. Getting rid of duplicate songs in a collection
2. Correctly tagging songs in an mp3 collection after the fact (ala musicbrainz)
3. P2P - get better quality search results.
4. Sharing metadata (other than id3) such as favourite songs number of times played etc. This info can enable us to find out lists of the most popular songs at the moment - etc.

Fingerprinting should be open, I am very surprised that the people researching this have not released a practical and accurate open engine for identifying songs.

Slashdot Top Deals

Biology is the only science in which multiplication means the same thing as division.

Working...