Single Channel Inverse Filtering of Room Impulse Response by Maximizing Skewness of LP Residual

This paper presents a single-microphone inverse filtering of room impulse response (RIR). The estimation is carried out by maximizing the skewness of the Linear Predictive (LP) residual of reverberant speech. The proposed algorithm used LP residual skewness to develop an on-line adaptive gradient-descent algorithm for calculating an inverse filter of RIR and proposed an affective technique to reduce amount of data size required for estimation. The comparison between our proposed method and the previous method based on one microphone shows that, it is more effective even in high reverberation time (upper than 1sec).