Analysis of high frequency partials in Bayesian harmonic model

Bayesian harmonic modeling and parameters estimation is a new approach in audio signal synthesizing. However current Bayesian harmonic modeling canpsilat effectively extract high frequency partials. In this paper we propose an improved model with parameters of high frequency partials for audio signal with harmonic modeling. We estimate partials in a Bayesian framework with the prior knowledge and likelihood function of the model parameters, then we use Monte Carlo Markov Chain (MCMC) sampling algorithm to approximate the posteriori distribution of the parameters. Our simulation shows that the new model and estimations can greatly improve the sound quality of the reconstructed audio signals.