Realtime Classification Of Hand-Drum Strokes
Krzyzaniak, Michael, Garth Paine · Zenodo (CERN European Organization for Nuclear Research) · 2015
Herein is presented a method of classifying hand-drum strokes in real-time by analyzing 50 milliseconds of audio signal as recorded by a contact-mic affixed to the body of the instrument. The classifier performs with an average accuracy of about 95% across several experiments on archetypical strokes, and 89% on uncontrived playing. A complete ANSI C implementation for OSX and Linux is available on the author's website.