The Kalman filtering concept has been widely used for control and filtering purposes in diverse fields, e.g. speech enhancement, navigation, etc. This paper investigates its application for enhancement and restoration of noisy audio signals. The topics of signal framing, signal modeling, and noise characterization, as well as issues of calculation burden and real-time implementation are discussed. Results of computer simulation as well as of hearing tests are presented and the performance of this concept is compared with that of other noise reduction methods.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.