Intelligent music accompaniment system based on discrete wavelet transform

Abstract

Instrument players usually play music with an unstable tempo. However, most of the robotic accompaniment systems are designed with a fixed beat. They can’t adjust their tempo to players’ tempo automatically. Therefore, we propose an adaptive accompaniment system for a soloist that can follow the tempo of a guitar player. The system includes a sound signal receiving element, an audio processing module, a robotic drumming system, and an information display device. The guitar signal is processed with the discrete wavelet transform, audio feature extraction, and peak detection to estimate the tempo of a piece of music, also known as the Beats Per Minute (BPM). In this research, the Daubechies Wavelet (db) is used for signal extraction and reconstruction. The experimental results validate wavelet five (db5) was the effective signal processing method. At the same time, the pedals are also designed to allow the guitarist to change the combination of bass drum and snare drum according to the performance needs. In the research, the guitar signal can be efficiently processed by Discrete Wavelet Transform and then control the drum to follow the guitarist’s tempo.