Invariant-integration method for robust feature extraction in speaker-independent speech recognition

Florian Müller; Alfred Mertins

Invariant-integration method for robust feature extraction in speaker-independent speech recognition

Institute of Signal Processing

Abstract

The vocal tract length (VTL) is one of the variabilities that speaker-independent automatic speech recognition (ASR) systems encounter. Standard methods to compensate for the effects of different VTLs within the processing stages of the ASR systems often have a high computational effort. By using an appropriate warping scheme for the frequency centers of the time-frequency analysis, a change in VTL can be approximately described by a translation in the subband-index space. We present a new type of features that is based on the principle of invariant integration, and an according feature selection method is described. ASR experiments show the increased robustness of the proposed features in comparison to standard MFCCs.

Original language	English
Title of host publication	Proc. Interspeech 2009
Number of pages	4
Publisher	International Speech Communication Association (ISCA)
Publication date	01.09.2009
Pages	2975-2978
Publication status	Published - 01.09.2009
Event	10th Annual Conference of the International Speech Communication Association - Brighton, United Kingdom Duration: 06.09.2009 → 10.09.2009 Conference number: 78744

UN SDGs

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

Access to Document

https://www.isca-speech.org/archive/archive_papers/interspeech_2009/papers/i09_2975.pdf

Cite this

Müller, F., & Mertins, A. (2009). Invariant-integration method for robust feature extraction in speaker-independent speech recognition. In Proc. Interspeech 2009 (pp. 2975-2978). (Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH). International Speech Communication Association (ISCA). https://www.isca-speech.org/archive/archive_papers/interspeech_2009/papers/i09_2975.pdf

@inproceedings{662df78813dd4efeb625d2c877c1ee04,

title = "Invariant-integration method for robust feature extraction in speaker-independent speech recognition",

abstract = "The vocal tract length (VTL) is one of the variabilities that speaker-independent automatic speech recognition (ASR) systems encounter. Standard methods to compensate for the effects of different VTLs within the processing stages of the ASR systems often have a high computational effort. By using an appropriate warping scheme for the frequency centers of the time-frequency analysis, a change in VTL can be approximately described by a translation in the subband-index space. We present a new type of features that is based on the principle of invariant integration, and an according feature selection method is described. ASR experiments show the increased robustness of the proposed features in comparison to standard MFCCs.",

author = "Florian M{\"u}ller and Alfred Mertins",

year = "2009",

month = sep,

day = "1",

language = "English",

series = "Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH",

publisher = " International Speech Communication Association (ISCA)",

pages = "2975--2978",

booktitle = "Proc. Interspeech 2009",

note = "10th Annual Conference of the International Speech Communication Association, INTERSPEECH 2009 ; Conference date: 06-09-2009 Through 10-09-2009",

}

Müller, F & Mertins, A 2009, Invariant-integration method for robust feature extraction in speaker-independent speech recognition. in Proc. Interspeech 2009. Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH, International Speech Communication Association (ISCA), pp. 2975-2978, 10th Annual Conference of the International Speech Communication Association, Brighton, United Kingdom, 06.09.09. <https://www.isca-speech.org/archive/archive_papers/interspeech_2009/papers/i09_2975.pdf>

Invariant-integration method for robust feature extraction in speaker-independent speech recognition. / Müller, Florian; Mertins, Alfred.
Proc. Interspeech 2009. International Speech Communication Association (ISCA), 2009. p. 2975-2978 (Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH).

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

TY - GEN

T1 - Invariant-integration method for robust feature extraction in speaker-independent speech recognition

AU - Müller, Florian

AU - Mertins, Alfred

N1 - Conference code: 78744

PY - 2009/9/1

Y1 - 2009/9/1

N2 - The vocal tract length (VTL) is one of the variabilities that speaker-independent automatic speech recognition (ASR) systems encounter. Standard methods to compensate for the effects of different VTLs within the processing stages of the ASR systems often have a high computational effort. By using an appropriate warping scheme for the frequency centers of the time-frequency analysis, a change in VTL can be approximately described by a translation in the subband-index space. We present a new type of features that is based on the principle of invariant integration, and an according feature selection method is described. ASR experiments show the increased robustness of the proposed features in comparison to standard MFCCs.

AB - The vocal tract length (VTL) is one of the variabilities that speaker-independent automatic speech recognition (ASR) systems encounter. Standard methods to compensate for the effects of different VTLs within the processing stages of the ASR systems often have a high computational effort. By using an appropriate warping scheme for the frequency centers of the time-frequency analysis, a change in VTL can be approximately described by a translation in the subband-index space. We present a new type of features that is based on the principle of invariant integration, and an according feature selection method is described. ASR experiments show the increased robustness of the proposed features in comparison to standard MFCCs.

UR - https://www.scopus.com/record/display.uri?eid=2-s2.0-70450220336&origin=inward&txGid=31f5522247b6517bbce3337a8ba7656d

M3 - Conference contribution

T3 - Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH

SP - 2975

EP - 2978

BT - Proc. Interspeech 2009

PB - International Speech Communication Association (ISCA)

T2 - 10th Annual Conference of the International Speech Communication Association

Y2 - 6 September 2009 through 10 September 2009

ER -

Invariant-integration method for robust feature extraction in speaker-independent speech recognition

Abstract

UN SDGs

Access to Document

Other files and links

Fingerprint

Invariant features for automatic speech recognition

Cite this

Invariant-integration method for robust feature extraction in speaker-independent speech recognition

Abstract

UN SDGs

Access to Document

Other files and links

Fingerprint

Projects

Invariant features for automatic speech recognition

Cite this