Transient detection of audio signals based on an adaptive comb filter in the frequency domain

This paper presents a transient detection algorithm suitable for rhythm detection in music signals. In many audio signals, low energy transients are masked by high energy stationary sounds. These masked transients, as well as higher energy and more visible transients, convey important information on the rhythm and time segmentation of the music signal. The proposed segmentation algorithm uses a sinusoidal model combined with adaptive comb filtering in the frequency domain to remove the stationary component of a sound signal. After filtering, the time envelope of the residual signal is analyzed to locate the transient components. Results show that the proposed algorithm can accurately detect most low energy transients.