AN EFFECTIVE KALMAN FILTERING METHOD FOR ENHANCING SPEECH IN THE PRESENCE OF COLORED NOISE

The Kalman filtering algorithm for speech enhancement is easily implemented and is efficient under white noise environments. This paper proposes an effective Kalman filtering algorithm for enhancing speech corrupted by colored noise, based on a whitened matrix. Compared with the conventional Kalman filtering algorithms to handle colored noise, the proposed Kalman filtering algorithm has low computational complexity and overcomes the difficulty of estimating the covariance matrix of colored noise. Simulation results confirm that the proposed Kalman filtering algorithm has better performance than several conventional algorithms in decreasing colored noise and speech distortion.