The effect of reverberation on the performance of cepstral mean subtraction in speaker verification

Noam R. Shabtai; Boaz Rafaely; Yaniv Zigel

doi:10.1016/j.apacoust.2010.09.009

The effect of reverberation on the performance of cepstral mean subtraction in speaker verification

Noam R. Shabtai, Boaz Rafaely, Yaniv Zigel

Research output: Contribution to journal › Article › peer-review

5 Scopus citations

Abstract

Speaker verification (SVR) performance is degraded under reverberation conditions. Cepstral mean subtraction (CMS) is often applied to the feature vectors in order to compensate for convolutive effects of transmission channels, which are considered to have a short-duration impulse response. The effect of reverberation on the performance of CMS applied to the feature vectors in SVR is investigated. Although CMS was found effective in reducing the effect of reverberation for short reverberation time (RT), in cases of long RT, it is shown that CMS may degrade SVR performance rather than improve it. Hence, CMS should not to be used in these cases. In addition, the effect of the room volume was tested and found less critical than the effect of long RT.

Original language	English
Pages (from-to)	124-126
Number of pages	3
Journal	Applied Acoustics
Volume	72
Issue number	2-3
DOIs	https://doi.org/10.1016/j.apacoust.2010.09.009
State	Published - 1 Feb 2011

Keywords

Cepstral mean subtraction
Reverberation
Speaker recognition

ASJC Scopus subject areas

Acoustics and Ultrasonics

Access to Document

10.1016/j.apacoust.2010.09.009

Cite this

@article{166441fdd5a44a3ab806d8e79ecdbf0a,

title = "The effect of reverberation on the performance of cepstral mean subtraction in speaker verification",

abstract = "Speaker verification (SVR) performance is degraded under reverberation conditions. Cepstral mean subtraction (CMS) is often applied to the feature vectors in order to compensate for convolutive effects of transmission channels, which are considered to have a short-duration impulse response. The effect of reverberation on the performance of CMS applied to the feature vectors in SVR is investigated. Although CMS was found effective in reducing the effect of reverberation for short reverberation time (RT), in cases of long RT, it is shown that CMS may degrade SVR performance rather than improve it. Hence, CMS should not to be used in these cases. In addition, the effect of the room volume was tested and found less critical than the effect of long RT.",

keywords = "Cepstral mean subtraction, Reverberation, Speaker recognition",

author = "Shabtai, {Noam R.} and Boaz Rafaely and Yaniv Zigel",

year = "2011",

month = feb,

day = "1",

doi = "10.1016/j.apacoust.2010.09.009",

language = "English",

volume = "72",

pages = "124--126",

journal = "Applied Acoustics",

issn = "0003-682X",

publisher = "Elsevier Ltd.",

number = "2-3",

}

TY - JOUR

T1 - The effect of reverberation on the performance of cepstral mean subtraction in speaker verification

AU - Shabtai, Noam R.

AU - Rafaely, Boaz

AU - Zigel, Yaniv

PY - 2011/2/1

Y1 - 2011/2/1

N2 - Speaker verification (SVR) performance is degraded under reverberation conditions. Cepstral mean subtraction (CMS) is often applied to the feature vectors in order to compensate for convolutive effects of transmission channels, which are considered to have a short-duration impulse response. The effect of reverberation on the performance of CMS applied to the feature vectors in SVR is investigated. Although CMS was found effective in reducing the effect of reverberation for short reverberation time (RT), in cases of long RT, it is shown that CMS may degrade SVR performance rather than improve it. Hence, CMS should not to be used in these cases. In addition, the effect of the room volume was tested and found less critical than the effect of long RT.

AB - Speaker verification (SVR) performance is degraded under reverberation conditions. Cepstral mean subtraction (CMS) is often applied to the feature vectors in order to compensate for convolutive effects of transmission channels, which are considered to have a short-duration impulse response. The effect of reverberation on the performance of CMS applied to the feature vectors in SVR is investigated. Although CMS was found effective in reducing the effect of reverberation for short reverberation time (RT), in cases of long RT, it is shown that CMS may degrade SVR performance rather than improve it. Hence, CMS should not to be used in these cases. In addition, the effect of the room volume was tested and found less critical than the effect of long RT.

KW - Cepstral mean subtraction

KW - Reverberation

KW - Speaker recognition

UR - http://www.scopus.com/inward/record.url?scp=78649463374&partnerID=8YFLogxK

U2 - 10.1016/j.apacoust.2010.09.009

DO - 10.1016/j.apacoust.2010.09.009

M3 - Article

AN - SCOPUS:78649463374

SN - 0003-682X

VL - 72

SP - 124

EP - 126

JO - Applied Acoustics

JF - Applied Acoustics

IS - 2-3

ER -

The effect of reverberation on the performance of cepstral mean subtraction in speaker verification

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this