a novel algorithm to segment an audio piece into its structural components. The boundaries of the homogeneous regions are decided based on various time and frequency domain features. The algorithm has been designed in 2 stages. In the first