Audio Analysis with Python and Vamp

Feeling optimistic about Aubio, via Sonic Annotator returning useful — if not all the details we need — to replace the Echonest Analysis API in our Infinite Glitch project. So we may not need to roll our own Vamp plugin and the fact that Sonic Annotator isn’t seeing the Vampy plugins on OS X Yosemite may not hamper our progress. It’s been two days since I signed up for the Vamp forum and still no confirmation email.

At any rate. Data can be returned in various formats. For example:

sonic-annotator -d vamp:vamp-aubio:aubiotempo:beats audio/77\ Williams\,\ Mars_id62.mp3 -w csv --csv-stdout
sonic-annotator -d vamp:vamp-aubio:aubiotempo:beats audio/77\ Williams\,\ Mars_id62.mp3 -w rdf --rdf-stdout

First thing I need to find are the locations in the current code that need the data. There are at least two features which require the audio analysis data: adding a new track and building a transition between tracks. I guess also the recording studio requires it to assemble the background and vocal tracks. I’ll run the app here locally and start poking around and looking at the stdout for clues.

While looking at that I decided to make another quick search and found a new replacement for EchoNest Remix! It’s called Amen, after the ubiquitous Amen Break.

Leave a Reply

Your email address will not be published. Required fields are marked *