By Gerard Roma
Uses the Freesound Website. www.freesound.org. The sounds are Creative Commons. The website has more than 150,000 sounds from around 4000 users. Most users only download sounds. All sounds are moderated – listened to by a human.
I’m always charmed when a presenter shows a supercollider window rather than using a slide programme. The syntax highlighting of their talk notes is especially good.
Google gave them a grant and they re-wrote the site. They have a feature extraction library to analyse the sounds.
There is a new freesound quark based on their API. The quark will give you the sound, the sound’s preview, the tags, the spectrogram, the signal descriptors from freesound’s feature extraction.
You need to get an API key to use the quark. The quark will search stuff for you according to filters. You can find a sound that’s glitchy with a particular duration. You can search by similarity as well.
The analysis frames of the sound are kept in a separate file, but can be loaded into an IdentityDictionary.
This quark could be really interesting if you want to do stuff with freesound, you don’t need to do your own MIR and you might be able to make cool pieces in real time.
Questions
- Are people doing cool things with this outside of SuperCollider? He doesn’t know.
- Will the API upload to freesound? No. The API needs some more authentication stuff put in. Also the moderation creates a delay.
- Zlatko wants to know about how they know if sounds are copyrighted. The moderators try to figure it out and respond to complaints.
- Can the same API key be used across multiple computers? Yes.
- Does the metadata include the licence terms and the user who uploaded it? Yes
- Is there a GUI? No, this is a new quark, not the old one.