We present a method for restoration of audio which has been distorted by a nonlinear channel or recording medium. We model the signal as a linear process cascaded with a nonlinear distortion. We use sampling methods to perform model selection and parameter estimation. We allow the linear model to be nonstationary, and present results for both synthetic and real distortions.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.