A comparison of front-end configurations for robust speech recognition

B. P. Milner

doi:10.1109/ICASSP.2002.5743838

A comparison of front-end configurations for robust speech recognition

B. P. Milner

Research output: Contribution to conference › Other

45 Citations (Scopus)

Abstract

This paper presents a comparative analysis of the processing stages involved in feature extraction for speech recognition. Feature extraction is considered as comprising three different processing stages; namely static feature extraction, normalisation and inclusion of temporal information. In each stage a comparison of techniques is made, both theoretically and in terms of their comparative performance. The analysis shows that while some techniques may appear significantly different, upon analysis the effect they have on the signal can be similar. Comparative studies include MFCC and PLP analysis, RASTA filtering and cepstral mean normalisation, and temporal derivatives and cepstral-time matrices. Experimental results, on an unconstrained monophone task, compare recognition performance using different front-end configurations.

Original language	English
DOIs	https://doi.org/10.1109/ICASSP.2002.5743838
Publication status	Published - May 2002
Event	IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP) - Minneapolis, United States Duration: 13 May 2002 → 17 May 2002

Conference

Conference	IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP)
Country/Territory	United States
City	Minneapolis
Period	13/05/02 → 17/05/02

Access to Document

10.1109/ICASSP.2002.5743838

Cite this

@conference{d504d523ca6e41cdae1a24704ff50bf2,

title = "A comparison of front-end configurations for robust speech recognition",

abstract = "This paper presents a comparative analysis of the processing stages involved in feature extraction for speech recognition. Feature extraction is considered as comprising three different processing stages; namely static feature extraction, normalisation and inclusion of temporal information. In each stage a comparison of techniques is made, both theoretically and in terms of their comparative performance. The analysis shows that while some techniques may appear significantly different, upon analysis the effect they have on the signal can be similar. Comparative studies include MFCC and PLP analysis, RASTA filtering and cepstral mean normalisation, and temporal derivatives and cepstral-time matrices. Experimental results, on an unconstrained monophone task, compare recognition performance using different front-end configurations.",

author = "Milner, {B. P.}",

year = "2002",

month = may,

doi = "10.1109/ICASSP.2002.5743838",

language = "English",

note = "IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP) ; Conference date: 13-05-2002 Through 17-05-2002",

}

TY - CONF

T1 - A comparison of front-end configurations for robust speech recognition

AU - Milner, B. P.

PY - 2002/5

Y1 - 2002/5

N2 - This paper presents a comparative analysis of the processing stages involved in feature extraction for speech recognition. Feature extraction is considered as comprising three different processing stages; namely static feature extraction, normalisation and inclusion of temporal information. In each stage a comparison of techniques is made, both theoretically and in terms of their comparative performance. The analysis shows that while some techniques may appear significantly different, upon analysis the effect they have on the signal can be similar. Comparative studies include MFCC and PLP analysis, RASTA filtering and cepstral mean normalisation, and temporal derivatives and cepstral-time matrices. Experimental results, on an unconstrained monophone task, compare recognition performance using different front-end configurations.

AB - This paper presents a comparative analysis of the processing stages involved in feature extraction for speech recognition. Feature extraction is considered as comprising three different processing stages; namely static feature extraction, normalisation and inclusion of temporal information. In each stage a comparison of techniques is made, both theoretically and in terms of their comparative performance. The analysis shows that while some techniques may appear significantly different, upon analysis the effect they have on the signal can be similar. Comparative studies include MFCC and PLP analysis, RASTA filtering and cepstral mean normalisation, and temporal derivatives and cepstral-time matrices. Experimental results, on an unconstrained monophone task, compare recognition performance using different front-end configurations.

U2 - 10.1109/ICASSP.2002.5743838

DO - 10.1109/ICASSP.2002.5743838

M3 - Other

T2 - IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP)

Y2 - 13 May 2002 through 17 May 2002

ER -