Retrieval of Perucssive Gesutres Using Indirect Acquisition
Adam Tindale & Ajay Kapur
McGill University and University of Victoria
The ability for a computer to listen to music and extract parameters is an ongoing research area. A musician can listen to music and instantly identify different instruments and the timbres produced by different playing techniques. This project uses digital signal processing machine learning techniques to
This project aims to use audio signals to determine when the snare drum has been hit and where it was struck. By developing techniques to allow the computer to extract percussion instruments from signals, we move closer to the day when the computer can listen to any signal and recognize it. Adapting tools that are already available in similar tasks will facilitate this research. These analysis techniques from these tasks will not only help lead to the completion of this project but provide tools for other researchers and musicians.
The study has begun with only five classification categories for the signal: rimshot, brush stroke, regular stroke in the center of the drum, regular stroke at the edge of the drum, regular stroke halfway between the edge and the center of the drum. As the study continues, it is hoped that it will be possible to increase the resolution of the snare drum tracking. Future experiments will be run with different striking force, smaller position resolutions, and different snare drummers.
For more information refer to the following articles: