Non-Intrusive Intelligibility Prediction Using a Codebook-Based Approach

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

1 Citation (Scopus)

Abstract

It could be beneficial for users of hearing aids if
these were able to automatically adjust the processing according
to the speech intelligibility in the specific acoustic environment.
Most speech intelligibility metrics are intrusive, i.e., they require
a clean reference signal, which is rarely available in real-life
applications. This paper proposes a method, which allows using
an intrusive short-time objective intelligibility (STOI) metric
without requiring access to a clean signal. The clean speech
reference signal is replaced by the clean speech envelope spectrum
estimated from the noisy signal. The spectral envelope has been
shown to be an important cue for speech intelligibility and is used
as the reference signal inside STOI. The spectral envelopes are
estimated as a combination of predefined dictionaries, i.e., codebooks,
that best fits the noisy speech signal. The simulations show
a high correlation between the proposed non-intrusive codebookbased
STOI (NIC-STOI) and the intrusive STOI indicating that
NIC-STOI is a suitable metric for automatic classification of
speech signals
Original languageEnglish
Title of host publication25th European Signal Processing Conference (EUSIPCO), 2017
PublisherIEEE
Publication date2017
Pages216-220
ISBN (Electronic)978-0-9928626-7-1
DOIs
Publication statusPublished - 2017
Event25th European Signal Processing Conference 2017 - Kos International Convention Center, Kos, Greece
Duration: 28 Aug 20172 Sep 2017
Conference number: 25
https://www.eusipco2017.org/#

Conference

Conference25th European Signal Processing Conference 2017
Number25
LocationKos International Convention Center
CountryGreece
CityKos
Period28/08/201702/09/2017
Internet address
SeriesProceedings of the European Signal Processing Conference
ISSN2076-1465

Fingerprint

Speech intelligibility
Hearing aids
Glossaries
Acoustics
Processing

Cite this

Sørensen, C., Kavalekalam, M. S., Xenaki, A., Boldt, J. B., & Christensen, M. G. (2017). Non-Intrusive Intelligibility Prediction Using a Codebook-Based Approach. In 25th European Signal Processing Conference (EUSIPCO), 2017 (pp. 216-220). IEEE. Proceedings of the European Signal Processing Conference https://doi.org/10.23919/EUSIPCO.2017.8081200
Sørensen, Charlotte ; Kavalekalam, Mathew Shaji ; Xenaki, Angeliki ; Boldt, Jesper Bünsow ; Christensen, Mads Græsbøll. / Non-Intrusive Intelligibility Prediction Using a Codebook-Based Approach. 25th European Signal Processing Conference (EUSIPCO), 2017. IEEE, 2017. pp. 216-220 (Proceedings of the European Signal Processing Conference).
@inproceedings{9d306e08aaf54e82b5191bf61bee3722,
title = "Non-Intrusive Intelligibility Prediction Using a Codebook-Based Approach",
abstract = "It could be beneficial for users of hearing aids ifthese were able to automatically adjust the processing accordingto the speech intelligibility in the specific acoustic environment.Most speech intelligibility metrics are intrusive, i.e., they requirea clean reference signal, which is rarely available in real-lifeapplications. This paper proposes a method, which allows usingan intrusive short-time objective intelligibility (STOI) metricwithout requiring access to a clean signal. The clean speechreference signal is replaced by the clean speech envelope spectrumestimated from the noisy signal. The spectral envelope has beenshown to be an important cue for speech intelligibility and is usedas the reference signal inside STOI. The spectral envelopes areestimated as a combination of predefined dictionaries, i.e., codebooks,that best fits the noisy speech signal. The simulations showa high correlation between the proposed non-intrusive codebookbasedSTOI (NIC-STOI) and the intrusive STOI indicating thatNIC-STOI is a suitable metric for automatic classification ofspeech signals",
author = "Charlotte S{\o}rensen and Kavalekalam, {Mathew Shaji} and Angeliki Xenaki and Boldt, {Jesper B{\"u}nsow} and Christensen, {Mads Gr{\ae}sb{\o}ll}",
year = "2017",
doi = "10.23919/EUSIPCO.2017.8081200",
language = "English",
pages = "216--220",
booktitle = "25th European Signal Processing Conference (EUSIPCO), 2017",
publisher = "IEEE",
address = "United States",

}

Sørensen, C, Kavalekalam, MS, Xenaki, A, Boldt, JB & Christensen, MG 2017, Non-Intrusive Intelligibility Prediction Using a Codebook-Based Approach. in 25th European Signal Processing Conference (EUSIPCO), 2017. IEEE, Proceedings of the European Signal Processing Conference, pp. 216-220, 25th European Signal Processing Conference 2017, Kos, Greece, 28/08/2017. https://doi.org/10.23919/EUSIPCO.2017.8081200

Non-Intrusive Intelligibility Prediction Using a Codebook-Based Approach. / Sørensen, Charlotte; Kavalekalam, Mathew Shaji; Xenaki, Angeliki; Boldt, Jesper Bünsow; Christensen, Mads Græsbøll.

25th European Signal Processing Conference (EUSIPCO), 2017. IEEE, 2017. p. 216-220.

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

TY - GEN

T1 - Non-Intrusive Intelligibility Prediction Using a Codebook-Based Approach

AU - Sørensen, Charlotte

AU - Kavalekalam, Mathew Shaji

AU - Xenaki, Angeliki

AU - Boldt, Jesper Bünsow

AU - Christensen, Mads Græsbøll

PY - 2017

Y1 - 2017

N2 - It could be beneficial for users of hearing aids ifthese were able to automatically adjust the processing accordingto the speech intelligibility in the specific acoustic environment.Most speech intelligibility metrics are intrusive, i.e., they requirea clean reference signal, which is rarely available in real-lifeapplications. This paper proposes a method, which allows usingan intrusive short-time objective intelligibility (STOI) metricwithout requiring access to a clean signal. The clean speechreference signal is replaced by the clean speech envelope spectrumestimated from the noisy signal. The spectral envelope has beenshown to be an important cue for speech intelligibility and is usedas the reference signal inside STOI. The spectral envelopes areestimated as a combination of predefined dictionaries, i.e., codebooks,that best fits the noisy speech signal. The simulations showa high correlation between the proposed non-intrusive codebookbasedSTOI (NIC-STOI) and the intrusive STOI indicating thatNIC-STOI is a suitable metric for automatic classification ofspeech signals

AB - It could be beneficial for users of hearing aids ifthese were able to automatically adjust the processing accordingto the speech intelligibility in the specific acoustic environment.Most speech intelligibility metrics are intrusive, i.e., they requirea clean reference signal, which is rarely available in real-lifeapplications. This paper proposes a method, which allows usingan intrusive short-time objective intelligibility (STOI) metricwithout requiring access to a clean signal. The clean speechreference signal is replaced by the clean speech envelope spectrumestimated from the noisy signal. The spectral envelope has beenshown to be an important cue for speech intelligibility and is usedas the reference signal inside STOI. The spectral envelopes areestimated as a combination of predefined dictionaries, i.e., codebooks,that best fits the noisy speech signal. The simulations showa high correlation between the proposed non-intrusive codebookbasedSTOI (NIC-STOI) and the intrusive STOI indicating thatNIC-STOI is a suitable metric for automatic classification ofspeech signals

U2 - 10.23919/EUSIPCO.2017.8081200

DO - 10.23919/EUSIPCO.2017.8081200

M3 - Article in proceeding

SP - 216

EP - 220

BT - 25th European Signal Processing Conference (EUSIPCO), 2017

PB - IEEE

ER -

Sørensen C, Kavalekalam MS, Xenaki A, Boldt JB, Christensen MG. Non-Intrusive Intelligibility Prediction Using a Codebook-Based Approach. In 25th European Signal Processing Conference (EUSIPCO), 2017. IEEE. 2017. p. 216-220. (Proceedings of the European Signal Processing Conference). https://doi.org/10.23919/EUSIPCO.2017.8081200