timbreIDLib is a library of audio analysis externals for Pure Data. The classification external [timbreID] accepts arbitrary lists of audio features and attempts to find the best match between an input feature and previously stored training instances. The library can be used for a variety of real-time and non-real-time applications, including sound classification, sound searching, sound visualization, automatic segmenting, ordering of sounds by timbre, key and tempo estimation, and concatenative synthesis.
Download the latest examples package here
Video demo of timbre space example