Standard

The ID R&D System Description for Short-duration Speaker Verification Challenge 2021. / Alenin, Alexander; Okhotnikov, Anton; Makarov, Rostislav; Torgashov, Nikita; Shigabeev, Ilya; Simonchik, Konstantin.

2021. 2297-2301 Работа представлена на Interspeech 2021, Брно, Чехия.

Результаты исследований: Материалы конференций › материалы › Рецензирование

Harvard

Alenin, A, Okhotnikov, A, Makarov, R, Torgashov, N, Shigabeev, I & Simonchik, K 2021, 'The ID R&D System Description for Short-duration Speaker Verification Challenge 2021', Работа представлена на Interspeech 2021, Брно, Чехия, 30/08/21 - 3/09/21 стр. 2297-2301. https://doi.org/10.21437/interspeech.2021-1553, https://doi.org/10.21437/Interspeech.2021-1553

APA

Alenin, A., Okhotnikov, A., Makarov, R., Torgashov, N., Shigabeev, I., & Simonchik, K. (2021). The ID R&D System Description for Short-duration Speaker Verification Challenge 2021. 2297-2301. Работа представлена на Interspeech 2021, Брно, Чехия. https://doi.org/10.21437/interspeech.2021-1553, https://doi.org/10.21437/Interspeech.2021-1553

Vancouver

Alenin A, Okhotnikov A, Makarov R, Torgashov N, Shigabeev I, Simonchik K. The ID R&D System Description for Short-duration Speaker Verification Challenge 2021. 2021. Работа представлена на Interspeech 2021, Брно, Чехия. https://doi.org/10.21437/interspeech.2021-1553, https://doi.org/10.21437/Interspeech.2021-1553

Author

Alenin, Alexander ; Okhotnikov, Anton ; Makarov, Rostislav ; Torgashov, Nikita ; Shigabeev, Ilya ; Simonchik, Konstantin. / The ID R&D System Description for Short-duration Speaker Verification Challenge 2021. Работа представлена на Interspeech 2021, Брно, Чехия.5 стр.

BibTeX

@conference{045ac95b303a4d478d54a8c416c26274,
title = "The ID R&D System Description for Short-duration Speaker Verification Challenge 2021",
abstract = "This paper describes ID R&D team submission to the text- independent task of the Short-duration Speaker Verification (SdSV) Challenge 2021. The top performed system is a fu- sion of 9 Convolutional Neural Networks based on the ResNet architecture. Experiments{\textquoteright} results of optimal NN architecture search are shown. We also present and investigate the subnet- work approach to solve the auxiliary tasks such as gender or language detection. Verification scores refinement step using quality measurements of a trial pair allowed to further mini- mize the target metrics. A comparative analysis of all systems used in the fusion has been provided on the VoxCeleb-1 test set, SdSV-2021 development and evaluation sets. The final submis- sion achieves 0.69% EER and 0.0319 minDCF on the challenge evaluation set.",
keywords = "Speaker recognition, Speaker verification, cross- lingual speaker verification, SdSV Challenge 2021, Cross-lingual speaker verification",
author = "Alexander Alenin and Anton Okhotnikov and Rostislav Makarov and Nikita Torgashov and Ilya Shigabeev and Konstantin Simonchik",
year = "2021",
month = aug,
day = "30",
doi = "10.21437/interspeech.2021-1553",
language = "English",
pages = "2297--2301",
note = "Interspeech 2021, Interspeech 2021 ; Conference date: 30-08-2021 Through 03-09-2021",
url = "https://www.interspeech2021.org/",

}

RIS

TY - CONF

T1 - The ID R&D System Description for Short-duration Speaker Verification Challenge 2021

AU - Alenin, Alexander

AU - Okhotnikov, Anton

AU - Makarov, Rostislav

AU - Torgashov, Nikita

AU - Shigabeev, Ilya

AU - Simonchik, Konstantin

PY - 2021/8/30

Y1 - 2021/8/30

N2 - This paper describes ID R&D team submission to the text- independent task of the Short-duration Speaker Verification (SdSV) Challenge 2021. The top performed system is a fu- sion of 9 Convolutional Neural Networks based on the ResNet architecture. Experiments’ results of optimal NN architecture search are shown. We also present and investigate the subnet- work approach to solve the auxiliary tasks such as gender or language detection. Verification scores refinement step using quality measurements of a trial pair allowed to further mini- mize the target metrics. A comparative analysis of all systems used in the fusion has been provided on the VoxCeleb-1 test set, SdSV-2021 development and evaluation sets. The final submis- sion achieves 0.69% EER and 0.0319 minDCF on the challenge evaluation set.

AB - This paper describes ID R&D team submission to the text- independent task of the Short-duration Speaker Verification (SdSV) Challenge 2021. The top performed system is a fu- sion of 9 Convolutional Neural Networks based on the ResNet architecture. Experiments’ results of optimal NN architecture search are shown. We also present and investigate the subnet- work approach to solve the auxiliary tasks such as gender or language detection. Verification scores refinement step using quality measurements of a trial pair allowed to further mini- mize the target metrics. A comparative analysis of all systems used in the fusion has been provided on the VoxCeleb-1 test set, SdSV-2021 development and evaluation sets. The final submis- sion achieves 0.69% EER and 0.0319 minDCF on the challenge evaluation set.

KW - Speaker recognition

KW - Speaker verification

KW - cross- lingual speaker verification

KW - SdSV Challenge 2021

KW - Cross-lingual speaker verification

UR - http://www.scopus.com/inward/record.url?scp=85119188286&partnerID=8YFLogxK

UR - https://www.mendeley.com/catalogue/0d38456a-d03d-3451-a7e9-3000821e7c09/

U2 - 10.21437/interspeech.2021-1553

DO - 10.21437/interspeech.2021-1553

M3 - Paper

SP - 2297

EP - 2301

T2 - Interspeech 2021

Y2 - 30 August 2021 through 3 September 2021

ER -

ID: 86369686