Low Bit Rate Monophonic Music Coding Based on Bayesian Method

摘要

Bayesian harmonic modeling and parameters estimation are widely used in music signal processing. In this paper, we built a low bit rate monophonic music coding system based on Bayesian audio harmonic model. Firstly, we extract the model parameters using Bayesian estimation, and then adopt different coding methods for different characteristics of fundamental frequency, harmonics and other parameters separately. Experiment result demonstrates that our method not only can realize efficient coding of arbitrary music signals at low bit rate less than 8kbps but also can play music in real time on the network.