On the Impact of Quantization on Binaural MVDR Beamforming

Konferenz: Speech Communication - 12. ITG-Fachtagung Sprachkommunikation
05.10.2016 - 07.10.2016 in Paderborn, Deutschland

Tagungsband: ITG-Fb. 267: Speech Communication

Seiten: 5Sprache: EnglischTyp: PDF

Persönliche VDE-Mitglieder erhalten auf diesen Artikel 10% Rabatt

Autoren:
Amini, Jamal; Hendriks, Richard C.; Heusdens, Richard (Circuits and Systems (CAS) Group, Delft University of Technology, 2628 CD Delft, The Netherlands)
Guo, Meng (Oticon A/S, Kongebakken 9, 2765 Smørum, Denmark)
Jensen, Jesper (Oticon A/S, Kongebakken 9, 2765 Smørum, Denmark & Electronic Systems Department, Aalborg University, 9100 Aalborg, Denmark)

Inhalt:
Multi-microphone noise reduction algorithms in binaural hearing aids which cooperate through a wireless link have the potential to become of great importance in future hearing aid systems. However, limited transmission capacity of such devices necessitates the data compression of signals transmitted from one hearing aid to the contralateral one. In this paper we study the impact of quantization as a data compression scheme on the performance of the multi-microphone noise reduction algorithms. Using the binaural minimum variance distortionless response (BMVDR) beamformer as an illustration, we propose a quantization aware beamforming scheme which uses a modified cross power spectral density (CPSD) of the system noise including the quantization noise (QN). Moreover, several assumptions on the QN are investigated in the proposed method. Based on the output SNR, we compare different variations of the proposed method with the conventional BMVDR beamformer. The results confirm the improved performance of the proposed method.