Downloads: 111 | Views: 178

M.Tech / M.E / PhD Thesis | Electronics & Communication Engineering | India | Volume 3 Issue 5, May 2014

Musical Audio Beat Tracking using Hidden Markov Model

Lata Dhulekar | Dr. S. K. Shah

Abstract: The proposed method describes the beat tracking system of musical audio signals. Here; the beat tracking deals with the explicit modeling of non-beat states and estimates the time between consecutive beat events. The expected accuracy of the estimated beats is also provided additionally. To predict the accuracy of the beat estimate a k-nearest neighbor regression algorithm is used. A database of 222 musical signals of various genres is used to statistically evaluate the performance for beat tracking system. A new perspective for beat detection is presented which can be used to enhance the performance of the proposed algorithm and relates it with the human tapping.

Keywords: Beat-tracking, onset detection, k-nearest neighbor k-NN regression, Hidden Markov Model, Musical Audio Processing

Edition: Volume 3 Issue 5, May 2014,

Pages: 1054 - 1058