A noise suppression technique is necessary in a wide range of applications including mobile communication and speech recogunition systems. We have previously proposed a noise suppressor using lattice filter which can cancel a white noise from an observed signal. Unfortunately, many practical noises are not white, and hence the conventional noise suppressor is not available for the practical noises. In this paper, we propose a new adaptive algorithm used for the lattice filter to suppress a colored noise. The proposed algorithm can be directly derived from the conventional time recursive algorithm. To extract a speech from a speech mixed with colored noise, the lattice filter with the proposed algorithm gives a noise replica whose auto-correlation is close to the noise's one. Subtracting the noise replica from the observed noisy speech, we can obtain an extracted speech. Simulation results showed that the proposed noise suppressor can extract a speech from a speech mixed with a tunnel noise which is a colored noise recorded in a practical environment.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.