Comparison of LMS- and RLS-based adaptive algorithms for speech enhancement

In speech communication areas, such as hands-free mobile telephones, teleconferencing, room acoustics or long-distance telephony, we often come into a situation, where we have to face problems with noise. This article deals with the analysis of most important LMS- and RLS-based adaptive algorithms for noise cancellation in speech signals. The stress lays mainly on explanation of the methodology of each algorithm, their inherent computational complexity and the comparison between them. For testing purposes, a number of car-cockpit experiments were conducted under real conditions to reveal the "DSP" properties of all algorithms.