The SRI system for the NIST OpenSAD 2015 speech activity detection evaluation

In this paper, we present the SRI system submission to the NIST OpenSAD 2015 speech activity detection (SAD) evaluation. We present results on three different development databases that we created from the provided data. We present system-development results for feature normalization; for feature fu...

Descripción completa

Guardado en:

Detalles Bibliográficos
Publicado:	2016
Materias:	Channel degradation Noise robustness Speech activity detection Calibration Speech Speech communication Speech processing Testing Adaptive calibration Bottleneck features Channel bottlenecks Channel degradations Decision threshold Feature normalization Speech activity detections Speech recognition
Acceso en línea:	https://bibliotecadigital.exactas.uba.ar/collection/paper/document/paper_2308457X_v08-12-September-2016_n_p3673_Graciarena http://hdl.handle.net/20.500.12110/paper_2308457X_v08-12-September-2016_n_p3673_Graciarena
Aporte de:	Biblioteca Digital - Facultad de Ciencias Exactas y Naturales (UBA) de Universidad de Buenos Aires

id	paper:paper_2308457X_v08-12-September-2016_n_p3673_Graciarena
record_format	dspace
spelling	paper:paper_2308457X_v08-12-September-2016_n_p3673_Graciarena2023-06-08T16:35:30Z The SRI system for the NIST OpenSAD 2015 speech activity detection evaluation Channel degradation Noise robustness Speech activity detection Calibration Speech Speech communication Speech processing Testing Adaptive calibration Bottleneck features Channel bottlenecks Channel degradations Decision threshold Feature normalization Noise robustness Speech activity detections Speech recognition In this paper, we present the SRI system submission to the NIST OpenSAD 2015 speech activity detection (SAD) evaluation. We present results on three different development databases that we created from the provided data. We present system-development results for feature normalization; for feature fusion with acoustic, voicing, and channel bottleneck features; and finally for SAD bottleneck-feature fusion. We present a novel technique called test adaptive calibration, which is designed to improve decision-threshold selection for each test waveform. We present unsupervised test adaptation of the fusion component and describe its tight synergy to the test adaptive calibration component. Finally, we present results on the evaluation test data and show how the proposed techniques lead to significant gains on channels unseen during training. Copyright © 2016 ISCA. 2016 https://bibliotecadigital.exactas.uba.ar/collection/paper/document/paper_2308457X_v08-12-September-2016_n_p3673_Graciarena http://hdl.handle.net/20.500.12110/paper_2308457X_v08-12-September-2016_n_p3673_Graciarena
institution	Universidad de Buenos Aires
institution_str	I-28
repository_str	R-134
collection	Biblioteca Digital - Facultad de Ciencias Exactas y Naturales (UBA)
topic	Channel degradation Noise robustness Speech activity detection Calibration Speech Speech communication Speech processing Testing Adaptive calibration Bottleneck features Channel bottlenecks Channel degradations Decision threshold Feature normalization Noise robustness Speech activity detections Speech recognition
spellingShingle	Channel degradation Noise robustness Speech activity detection Calibration Speech Speech communication Speech processing Testing Adaptive calibration Bottleneck features Channel bottlenecks Channel degradations Decision threshold Feature normalization Noise robustness Speech activity detections Speech recognition The SRI system for the NIST OpenSAD 2015 speech activity detection evaluation
topic_facet	Channel degradation Noise robustness Speech activity detection Calibration Speech Speech communication Speech processing Testing Adaptive calibration Bottleneck features Channel bottlenecks Channel degradations Decision threshold Feature normalization Noise robustness Speech activity detections Speech recognition
description	In this paper, we present the SRI system submission to the NIST OpenSAD 2015 speech activity detection (SAD) evaluation. We present results on three different development databases that we created from the provided data. We present system-development results for feature normalization; for feature fusion with acoustic, voicing, and channel bottleneck features; and finally for SAD bottleneck-feature fusion. We present a novel technique called test adaptive calibration, which is designed to improve decision-threshold selection for each test waveform. We present unsupervised test adaptation of the fusion component and describe its tight synergy to the test adaptive calibration component. Finally, we present results on the evaluation test data and show how the proposed techniques lead to significant gains on channels unseen during training. Copyright © 2016 ISCA.
title	The SRI system for the NIST OpenSAD 2015 speech activity detection evaluation
title_short	The SRI system for the NIST OpenSAD 2015 speech activity detection evaluation
title_full	The SRI system for the NIST OpenSAD 2015 speech activity detection evaluation
title_fullStr	The SRI system for the NIST OpenSAD 2015 speech activity detection evaluation
title_full_unstemmed	The SRI system for the NIST OpenSAD 2015 speech activity detection evaluation
title_sort	sri system for the nist opensad 2015 speech activity detection evaluation
publishDate	2016
url	https://bibliotecadigital.exactas.uba.ar/collection/paper/document/paper_2308457X_v08-12-September-2016_n_p3673_Graciarena http://hdl.handle.net/20.500.12110/paper_2308457X_v08-12-September-2016_n_p3673_Graciarena
_version_	1768543775114657792

The SRI system for the NIST OpenSAD 2015 speech activity detection evaluation

Ejemplares similares