TY - JOUR

T1 - Blind compensation of interchannel sampling frequency mismatch for ad hoc microphone array based on maximum likelihood estimation

AU - Miyabe, Shigeki

AU - Ono, Nobutaka

AU - Makino, Shoji

N1 - Publisher Copyright:
© 2014 The Authors.

PY - 2015/2

Y1 - 2015/2

N2 - In this paper, we propose a novel method for the blind compensation of drift for the asynchronous recording of an ad hoc microphone array. Digital signals simultaneously observed by different recording devices have drift of the time differences between the observation channels because of the sampling frequency mismatch among the devices. On the basis of a model in which the time difference is constant within each short time frame but varies in proportion to the central time of the frame, the effect of the sampling frequency mismatch can be compensated in the short-time Fourier transform (STFT) domain by a linear phase shift. By assuming that the sources are motionless and have stationary amplitudes, the observation is regarded as being stationary when drift does not occur. Thus, we formulate a likelihood to evaluate the stationarity in the STFT domain to evaluate the compensation of drift. The maximum likelihood estimation is obtained effectively by a golden section search. Using the estimated parameters, we compensate the drift by STFT analysis with a noninteger frame shift. The effectiveness of the proposed blind drift compensation method is evaluated in an experiment in which artificial drift is generated.

AB - In this paper, we propose a novel method for the blind compensation of drift for the asynchronous recording of an ad hoc microphone array. Digital signals simultaneously observed by different recording devices have drift of the time differences between the observation channels because of the sampling frequency mismatch among the devices. On the basis of a model in which the time difference is constant within each short time frame but varies in proportion to the central time of the frame, the effect of the sampling frequency mismatch can be compensated in the short-time Fourier transform (STFT) domain by a linear phase shift. By assuming that the sources are motionless and have stationary amplitudes, the observation is regarded as being stationary when drift does not occur. Thus, we formulate a likelihood to evaluate the stationarity in the STFT domain to evaluate the compensation of drift. The maximum likelihood estimation is obtained effectively by a golden section search. Using the estimated parameters, we compensate the drift by STFT analysis with a noninteger frame shift. The effectiveness of the proposed blind drift compensation method is evaluated in an experiment in which artificial drift is generated.

KW - Ad hoc microphone array

KW - Blind source separation

KW - Drift

KW - Maximum likelihood estimation

KW - Sampling frequency

UR - http://www.scopus.com/inward/record.url?scp=84922448682&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84922448682&partnerID=8YFLogxK

U2 - 10.1016/j.sigpro.2014.09.015

DO - 10.1016/j.sigpro.2014.09.015

M3 - Article

AN - SCOPUS:84922448682

VL - 107

SP - 185

EP - 196

JO - Signal Processing

JF - Signal Processing

SN - 0165-1684

ER -