I'm looking for a programmer who can set up a REST web service which can match (part) of sounds to the total (original) sound. The technology would not be unlike what Shazam has. I have partial (but working) Java code and ofcourse the original whitepaper from Shazam and from some other companies who are doing the same thing.
The webservice should be able to do the following:
- upload a sample (different formats; .caf, .wav and mp3)
- check the sample against the sound fingerprint database
- send back a record from the database matching the sample in JSON (this record will contain all information about the 'artist' including images, sound bites , description, name, links)
Note; if you do not know Shazam (or one of their competitors) please check it before starting; we are particularly *not* looking for something which matches up 2 .sound files byte-for-byte. So please check that first. The algorithm and Java code you'll receive already works that way already, so you only have to finish, test and build the surrounding service/code.
Note; the algorithm will not be used for recognizing music, but for other long playing sounds. It has to be tested and maybe repurposed for that; this is up to you. Enough samples will be provided.
It would perhaps be best to cut the project in 2 parts where the first part is some research / testing to see if it will actually work with the kind of audio we are going to work with. I can make this a separate project or a milestone of this one.