SG-VAD: Stochastic Gates Based Speech Activity Detection

Jonathan Svirsky; Ofir Lindenbaum

doi:10.1109/icassp49357.2023.10096938

SG-VAD: Stochastic Gates Based Speech Activity Detection

Jonathan Svirsky, Ofir Lindenbaum

Research output: Contribution to journal › Conference article › peer-review

4 Scopus citations

Abstract

We propose a novel voice activity detection (VAD) model in a low-resource environment. Our key idea is to model VAD as a denoising task and construct a network that is designed to identify nuisance features for a speech classification task. We train the model to simultaneously identify irrelevant features while predicting the type of speech event. Our model contains only 7.8K parameters, outperforms the previously proposed methods on the AVA-Speech evaluation set, and provides comparative results on the HAVIC dataset. We present its architecture, experimental results, and ablation study on the model's components.

Original language	English
Journal	Proceedings - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing
DOIs	https://doi.org/10.1109/icassp49357.2023.10096938
State	Published - 2023
Event	48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023 - Rhodes Island, Greece Duration: 4 Jun 2023 → 10 Jun 2023

Bibliographical note

Publisher Copyright:
© 2023 IEEE.

Keywords

Feature Selection
Speech Recognition
Voice Activity Detection

Access to Document

10.1109/icassp49357.2023.10096938

Cite this

@article{3001e18d2bc54ad3922e10814d3e3c93,

title = "SG-VAD: Stochastic Gates Based Speech Activity Detection",

abstract = "We propose a novel voice activity detection (VAD) model in a low-resource environment. Our key idea is to model VAD as a denoising task and construct a network that is designed to identify nuisance features for a speech classification task. We train the model to simultaneously identify irrelevant features while predicting the type of speech event. Our model contains only 7.8K parameters, outperforms the previously proposed methods on the AVA-Speech evaluation set, and provides comparative results on the HAVIC dataset. We present its architecture, experimental results, and ablation study on the model's components.",

keywords = "Feature Selection, Speech Recognition, Voice Activity Detection",

author = "Jonathan Svirsky and Ofir Lindenbaum",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023 ; Conference date: 04-06-2023 Through 10-06-2023",

year = "2023",

doi = "10.1109/icassp49357.2023.10096938",

language = "אנגלית",

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

issn = "1520-6149",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - JOUR

T1 - SG-VAD

T2 - 48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023

AU - Svirsky, Jonathan

AU - Lindenbaum, Ofir

PY - 2023

Y1 - 2023

N2 - We propose a novel voice activity detection (VAD) model in a low-resource environment. Our key idea is to model VAD as a denoising task and construct a network that is designed to identify nuisance features for a speech classification task. We train the model to simultaneously identify irrelevant features while predicting the type of speech event. Our model contains only 7.8K parameters, outperforms the previously proposed methods on the AVA-Speech evaluation set, and provides comparative results on the HAVIC dataset. We present its architecture, experimental results, and ablation study on the model's components.

AB - We propose a novel voice activity detection (VAD) model in a low-resource environment. Our key idea is to model VAD as a denoising task and construct a network that is designed to identify nuisance features for a speech classification task. We train the model to simultaneously identify irrelevant features while predicting the type of speech event. Our model contains only 7.8K parameters, outperforms the previously proposed methods on the AVA-Speech evaluation set, and provides comparative results on the HAVIC dataset. We present its architecture, experimental results, and ablation study on the model's components.

KW - Feature Selection

KW - Speech Recognition

KW - Voice Activity Detection

UR - http://www.scopus.com/inward/record.url?scp=85180578746&partnerID=8YFLogxK

U2 - 10.1109/icassp49357.2023.10096938

DO - 10.1109/icassp49357.2023.10096938

M3 - ???researchoutput.researchoutputtypes.contributiontojournal.conferencearticle???

AN - SCOPUS:85180578746

SN - 1520-6149

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

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

Y2 - 4 June 2023 through 10 June 2023

ER -

SG-VAD: Stochastic Gates Based Speech Activity Detection

Abstract

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this