Extension of virtual microphone technique to multiple real microphones and investigation of the impact of phase and amplitude interpolation on speech enhancement

Hanako Segawa, Li Li, Shoji Makino, Takeshi Yamada

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

Abstract

A virtual microphone signal can artificially increase the number of channels of an observed signal, which leads to improved multi-channel speech enhancement performance. The conventional virtual microphone technique can only be applied to two real microphones, reducing its application to a limited range of situations. In this paper, we extend the virtual microphone technique to the interpolation of more than two real microphones. In the proposed method, the phase based on the plane wave model is linearly interpolated using an affine coefficients of the coordinates, and the amplitude is nonlinearly interpolated based on the β -divergence using coefficients inversely proportional to the distance between the virtual and real microphones. The experimental results shows that the proposed method is effective in improving speech enhancement performance. Furthermore, we investigate the impact of phase and amplitude interpolation on speech enhancement performance with two or three real microphones. The experimental results indicated that the benefits of phase interpolation based on plane wave model were limited, and that amplitude interpolation was important for improving speech enhancement performance.

Original languageEnglish
Title of host publication2021 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2021 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages597-602
Number of pages6
ISBN (Electronic)9789881476890
Publication statusPublished - 2021
Event2021 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2021 - Tokyo, Japan
Duration: 2021 Dec 142021 Dec 17

Publication series

Name2021 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2021 - Proceedings

Conference

Conference2021 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2021
Country/TerritoryJapan
CityTokyo
Period21/12/1421/12/17

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Vision and Pattern Recognition
  • Signal Processing
  • Instrumentation

Fingerprint

Dive into the research topics of 'Extension of virtual microphone technique to multiple real microphones and investigation of the impact of phase and amplitude interpolation on speech enhancement'. Together they form a unique fingerprint.

Cite this