Frequency-Warping Invariant Features for Automatic Speech Recognition

Alfred Mertins; Jan Rademacher

doi:10.1109/ICASSP.2006.1661453

Frequency-Warping Invariant Features for Automatic Speech Recognition

Alfred Mertins, Jan Rademacher

Institute of Signal Processing

University of Oldenburg

Abstract

Based on the well-known relationship between vocal tract length (VTL) variation and linear frequency warping, we present a method for generating vocal tract length invariant (VTLI) features. These features are computed as translation invariant, correlation-type features in a log-frequency domain. In phoneme classification and recognition experiments on the TIMIT database, their discrimination capabilities and robustness to mismatches between training and test conditions turned out to be considerably better than for Mel-frequency cepstral coefficients (MFCCs). The best results are obtained when VTLI features and MFCCs are combined

Original language	English
Title of host publication	2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings
Number of pages	4
Publisher	IEEE
Publication date	01.12.2006
Pages	1025-1028
Article number	1661453
ISBN (Print)	978-142440469-8
DOIs	https://doi.org/10.1109/ICASSP.2006.1661453
Publication status	Published - 01.12.2006
Event	2006 IEEE International Conference on Acoustics, Speech and Signal Processing - Toulouse, France Duration: 14.05.2006 → 19.05.2006 Conference number: 69350

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1109/ICASSP.2006.1661453

Cite this

Mertins, A., & Rademacher, J. (2006). Frequency-Warping Invariant Features for Automatic Speech Recognition. In 2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings (pp. 1025-1028). Article 1661453 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). IEEE. https://doi.org/10.1109/ICASSP.2006.1661453

@inproceedings{7569c3d18368423bb8b80efa3d30e5b6,

title = "Frequency-Warping Invariant Features for Automatic Speech Recognition",

abstract = "Based on the well-known relationship between vocal tract length (VTL) variation and linear frequency warping, we present a method for generating vocal tract length invariant (VTLI) features. These features are computed as translation invariant, correlation-type features in a log-frequency domain. In phoneme classification and recognition experiments on the TIMIT database, their discrimination capabilities and robustness to mismatches between training and test conditions turned out to be considerably better than for Mel-frequency cepstral coefficients (MFCCs). The best results are obtained when VTLI features and MFCCs are combined",

author = "Alfred Mertins and Jan Rademacher",

year = "2006",

month = dec,

day = "1",

doi = "10.1109/ICASSP.2006.1661453",

language = "English",

isbn = "978-142440469-8",

series = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",

publisher = "IEEE",

pages = "1025--1028",

booktitle = "2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings",

note = "2006 IEEE International Conference on Acoustics, Speech and Signal Processing , ICASSP 2006 ; Conference date: 14-05-2006 Through 19-05-2006",

}

Mertins, A & Rademacher, J 2006, Frequency-Warping Invariant Features for Automatic Speech Recognition. in 2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings., 1661453, ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, IEEE, pp. 1025-1028, 2006 IEEE International Conference on Acoustics, Speech and Signal Processing , Toulouse, France, 14.05.06. https://doi.org/10.1109/ICASSP.2006.1661453

Frequency-Warping Invariant Features for Automatic Speech Recognition. / Mertins, Alfred; Rademacher, Jan.
2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings. IEEE, 2006. p. 1025-1028 1661453 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings).

Research output: Chapters in Books/Reports/Conference Proceedings › Conference contribution › peer-review

TY - GEN

T1 - Frequency-Warping Invariant Features for Automatic Speech Recognition

AU - Mertins, Alfred

AU - Rademacher, Jan

N1 - Conference code: 69350

PY - 2006/12/1

Y1 - 2006/12/1

N2 - Based on the well-known relationship between vocal tract length (VTL) variation and linear frequency warping, we present a method for generating vocal tract length invariant (VTLI) features. These features are computed as translation invariant, correlation-type features in a log-frequency domain. In phoneme classification and recognition experiments on the TIMIT database, their discrimination capabilities and robustness to mismatches between training and test conditions turned out to be considerably better than for Mel-frequency cepstral coefficients (MFCCs). The best results are obtained when VTLI features and MFCCs are combined

AB - Based on the well-known relationship between vocal tract length (VTL) variation and linear frequency warping, we present a method for generating vocal tract length invariant (VTLI) features. These features are computed as translation invariant, correlation-type features in a log-frequency domain. In phoneme classification and recognition experiments on the TIMIT database, their discrimination capabilities and robustness to mismatches between training and test conditions turned out to be considerably better than for Mel-frequency cepstral coefficients (MFCCs). The best results are obtained when VTLI features and MFCCs are combined

UR - https://www.scopus.com/record/display.uri?eid=2-s2.0-33947666117&origin=inward&txGid=0f8328b0efc9346b78d2f975428c6881

U2 - 10.1109/ICASSP.2006.1661453

DO - 10.1109/ICASSP.2006.1661453

M3 - Conference contribution

SN - 978-142440469-8

T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

SP - 1025

EP - 1028

BT - 2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings

PB - IEEE

T2 - 2006 IEEE International Conference on Acoustics, Speech and Signal Processing

Y2 - 14 May 2006 through 19 May 2006

ER -

Frequency-Warping Invariant Features for Automatic Speech Recognition

Abstract

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this