Joint Bayesian removal of impulse and background noise

We present a method for the removal of noise including nonGaussian impulses from a signal. Impulse noise is removed jointly a homogenous Gaussian noise floor using a Gabor regression model [1]. The problem is formulated in a joint Bayesian framework and we use a Gibbs MCMC sampler to estimate parameters. We show how to deal with variable magnitude impulses using a shifted inverse gamma distribution for their variance. Our results show improved signal to noise ratios and perceived audio quality by explicitly modelling impulses with a discrete switching process and a new heavy-tailed amplitude model.

• Type: Journal
• Year: 2011
• Where: ICASSP
• Authors: James Murphy, Simon J. Godsill
• Keywords: Gabor Regression Model, Homogenous Gaussian Noise, ICASSP 2011, Impulses, Signal Processing