Title :
Joint Pitch and DOA Estimation Using the ESPRIT Method
Author :
Yuntao Wu ; Amir, Leshem ; Jensen, Jesper Rindom ; Guisheng Liao
Author_Institution :
Sch. of Comput. Sci. & Eng., Wuhan Inst. of Technol., Wuhan, China
Abstract :
In this paper, the problem of joint multi-pitch and direction-of-arrival (DOA) estimation for multichannel harmonic sinusoidal signals is considered. A spatio-temporal matrix signal model for a uniform linear array is defined, and then the ESPRIT method based on subspace techniques that exploits the invariance property in the time domain is first used to estimate the multi pitch frequencies of multiple harmonic signals. Followed by the estimated pitch frequencies, the DOA estimations based on the ESPRIT method are also presented by using the shift invariance structure in the spatial domain. Compared to the existing state-of-the-art algorithms, the proposed method based on ESPRIT without 2-D searching is computationally more efficient but performs similarly. An asymptotic performance analysis of the DOA and pitch estimation of the proposed method are also presented. Finally, the effectiveness of the proposed method is illustrated on a synthetic signal as well as real-life recorded data.
Keywords :
direction-of-arrival estimation; matrix algebra; time-domain analysis; DOA estimation; ESPRIT method; asymptotic performance analysis; direction-of-arrival estimation; invariance property; multichannel harmonic sinusoidal signals; multipitch frequencies; real-life recorded data; shift invariance structure; spatial domain; spatiotemporal matrix signal model; subspace techniques; synthetic signal; time domain; uniform linear array; Direction-of-arrival estimation; Estimation; Harmonic analysis; Joints; Speech; Speech processing; Vectors; DOA estimation; ESPRIT; performance analysis; pitch estimation;
Journal_Title :
Audio, Speech, and Language Processing, IEEE/ACM Transactions on
DOI :
10.1109/TASLP.2014.2367817