A new iterative speech enhancement scheme based on Kalman filtering

A new iterative speech enhancement scheme that can be seen as an approximation to the Expectation-Maximization (EM) algorithm is proposed. The algorithm employs a Kalman filter that models the excitation source as a spectrally white process with a rapidly time-varying variance, which calls for a high temporal resolution estimation of this variance. A Local Variance Estimator based on a Prediction Error Kalman Filter is designed for this high temporal resolution variance estimation. To achieve fast convergence and avoid local maxima of the likelihood function, a Weighted Power Spectral Subtraction filter is introduced as an initialization procedure. Iterations are then made sequential inter-frame, exploiting the fact that the AR model changes slowly between neighboring frames. The proposed algorithm is computationally more efficient than a baseline EM algorithm due to its fast convergence. Performance comparison shows significant improvement over the baseline EM algorithm in terms of three objective measures. Listening test indicates an improvement in subjective quality due to a significant reduction of musical noise compared to the baseline EM algorithm.