Evaluation of multichannel hearing aid system by rank-constrained spatial covariance matrix estimation

Masakazu Une, Yuki Kubo, Norihiro Takamune, Daichi Kitamura, Hiroshi Saruwatari, Shoji Makino

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In a noisy environment, speech extraction techniques make hearing aid systems more effective and practical. Blind source separation (BSS) is suitable for hearing aids because it can be employed without any a priori spatial information. Among many BSS methods, independent low-rank matrix analysis (ILRMA) achieves high-quality separation performance. In a diffuse-noise environment, however, ILRMA cannot suppress the noise since it is based on the determined situation. On the other hand, rank-constrained spacial covariance matrix (SCM) estimation overcomes the problem. This method utilizes spatial parameters accurately estimated by ILRMA and compensates for the deficiency of the spatial basis of diffuse noise. The application of BSS methods to a multichannel binaural hearing aid system with a smartphone has never been studied in detail thus far. To clarify the efficacy of the BSS methods in real environments, we record real sounds by constructing a hearing aid system with a dummy head and a smartphone. In this study, we investigate the applicability of BSS for a multichannel binaural hearing aid system with microphones on a smartphone. Furthermore, we apply ILRMA and the rank-constrained SCM estimation to the recorded data and evaluate these methods in terms of their separation performance.

Original languageEnglish
Title of host publication2019 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1874-1879
Number of pages6
ISBN (Electronic)9781728132488
DOIs
Publication statusPublished - 2019 Nov
Externally publishedYes
Event2019 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2019 - Lanzhou, China
Duration: 2019 Nov 182019 Nov 21

Publication series

Name2019 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2019

Conference

Conference2019 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2019
CountryChina
CityLanzhou
Period19/11/1819/11/21

ASJC Scopus subject areas

  • Information Systems

Fingerprint Dive into the research topics of 'Evaluation of multichannel hearing aid system by rank-constrained spatial covariance matrix estimation'. Together they form a unique fingerprint.

Cite this