Least-squares estimation of the common acoustical poles in room acoustics and head related transfer functions

Abstract

Accurate and stable estimation of common acoustical poles is required in many applications such as speech dereverberation, acoustic feedback cancellation, sound field equalization, and virtual auditory display. In this paper, an algorithm for the least-squares estimation of the acoustical poles is presented. In contrast to previous methods, which may have limited accuracy and/or limitations imposed by nonlinear optimization, our method precisely estimates the acoustical poles after a low number of iterations. The proposed method unconditionally guarantees stability of the estimated poles. The method is examined using measured head related and room transfer functions. Experimental results show that estimated common acoustical poles accurately match the resonance frequencies of the ear canal and precisely agree with theoretical poles for a room acoustic response.