Speech Reconstruction by Sparse Linear Prediction

This paper proposes a new variant of the least square autoregressive (LSAR) method for speech reconstruction, which can estimate via least squares a segment of missing samples by applying the linear prediction (LP) model of speech. First, we show that the use of a single high-order linear predictor can provide better results than the classic LSAR techniques based on shortand long-term predictors without the need of a pitch detector. However, this high-order predictor may reduce the reconstruction performance due to estimation errors, especially in the case of short pitch periods, and non-stationarity. In order to overcome these problems, we propose the use of a sparse linear predictor which resembles the classical speech model, based on shortand long-term correlations, where many LP coefficients are zero. The experimental results show the superiority of the proposed approach in both signal to noise ratio and perceptual performance.