Open source audio pattern recognition (finger printing) [closed]

Answer 1

If you are looking for a solution in .NET check SoundFingerprinting library.

It's open source and built on top of Content Fingerprinting Using Wavelets research paper.

The algorithm is different from Shazaam's, but the general idea is similar: extract most prominent coefficients from the spectrum, then use them to build the fingerprints for later retrieval.

Description of the algorithm can be found here.

Answer 2

ok. different search keywords, all the results I looked for.

http://musicbrainz.org/doc/Audio_Fingerprint

Thanks for anyone viewing

Answer 3

• echoprint looks like a solid alternative to commercial libs.

for speech recognition:

• CMUSphinx
• OpenEars

we only tested echoprint and so far so good.

I know the answer is a bit late but as this page pops up in google, we may as well improve it over time :)

edited as CMU is speech oriented

Answer 4

If you are looking for JAVA library then go for MusicG, I have used it in my one of past project and its working fine. http://code.google.com/p/musicg/

Answer 5

There are a few Open Source audio recognition projects out there, although you can forget the quality/performances of commercial audio identification services such as Shazam, SoundHound, etc. It also depends on what kind of recognition you want to do (audio files fingerprinting/tagging, real-time identification, OTA (Over-The-Air) recognition, etc.)

Open Surce

• AcoustID/MusicBrainz Provides tools to perform audio fingerprinting and metadata association (fingerprinter and server). You can put up your own fingerprint/metadata server or use MusicBrainz's service. Good solution for audio file fingerprinting and recognition, not suitable for real-time high-performances applications (i.e. OTA). We have also tested it for audio stream monitoring (broadcast monitoring) but the results were quite poor.

• Echoprint Provides the same tools as AcoustID (fingerprinter and server+metadata) and can also be used for audio stream monitoring as it can recognize snippets taken from anywhere within the audio with decent accuracy (but i wouldn't use it for serious commercial applications). They also claim it is suitable for OTA applications but the performances are nowhere suitable for production uses.

• Last.fm Not fully open source, although they have open sourced the fingerprinting module. You have to query their identification and metadata provider services as they do not provide a full fledged solution.