I'm an italian student of sound engineering. I'm trying to develop a small project of query-by-sample audio retrieval based on a fast indexing algorythm (Locality sensitive hashing, if you want to know).
I need a large database of audio files and freesound seems to be just perfect for my purposes. The only problem is that I need a way to automatically access to and download the uploaded files.
I was more than happy when I saw the Mootcher API, but reading here and there I understood that the project is abandoned and no more working.
Is there any other usable API or interface (in any language) that I can use?
Thanks in advance,
My best guess would be to ftp from freesound.org into your own databases. This would not only grant you the use of any language/compiler/os/arch etc. you please, but also allow you to have complete control from the ground up.
Concerning the amount of files that you'd have to download for any substantial research about LSH, it would have to be a lot (I'm sure you are aware of this, or you wouldn't know LSH in the first place!) A solution would be to programmatically download sound files from fs.org. The time an effort are worth it as there are 85,000+ sample, so any inferences about LSH, and more specifically data sets acquired when testing O notated algorithms, would be reliable (my opinion).
My idea was to programmatically download one sample at time, and for each one compute a feature vector, put it in a db and then delete the file from my hard drive. I'm trying to guess a quick way to login to freesound looking at the Jmootcher libraries (which are obviously not working).
I just need to know the exact HTML POST message to send in the page http://www.freesound.org/forum/login.php , the rest should be easy enough. Anyone skilled in web programming and PHP?
I would also love some help in using the freesound API.
To capture the POST header (and header traffic in general) you might use a firefox plugin like Firebug or Live HTTP Headers. But you should also note that to download files after logging in your client should maintain the session ID which is most probably stored in a cookie. I think that there are some Python packages that make building a simple HTTP client fairly easy. So if you happen to know Python that would be the way to go.
I was asking here http://forum.cockos.com/showthread.php?t=42678 to get Freesound integrated into REAPER. But one guy said it doesn't work:
Just tried to use the Mootcher C++ API. It doesn't work. Known problem, it's been out of development for a long time apparently. So anyone wanting to get access from their programs to Freesound must roll their own implementation.
Any hints or news?
Freesound 2.0 will have a full featured API. Keep an eye on the blog.
Eagerly looking forward to it. Thanks Mr. B!