Python toolkit for performance-invariant popular music fingerprinting and song description, as presented in [1]. Intended for applications in music analysis and large-scale cover song retrieval.
NOTE: Please refer to the following repositories for a better / more recent implemenation of some of the fingerprinting methods in this package:
cover_id : Implements fingerprint learning as described in [1].
catchy : Implements more tools for the corpus analysis, including some of the pitch features described in [1].
Imports Numpy and Scipy, and optionally Pandas for i/o.
[1] Van Balen, J., Wiering, F., & Veltkamp, R. (2015). Audio Bigrams as a Unifying Model of Pitch-based Song Description. 11th International Symposium on Computer Music Multidisciplinary Research (CMMR).
Home page: http://www.github.com/jvbalen/pytch
© 2014 Jan Van Balen (@jvanbalen on Twitter)