Attention-Based Temporal-Frequency Aggregation for Speaker Verification
Convolutional neural networks (CNNs) have significantly promoted the development of speaker verification (SV) systems because of their powerful deep feature learning capability. In CNN-based SV systems, utterance-level aggregation is an important component, and it compresses the frame-level features...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2022-03-01
|
Series: | Sensors |
Subjects: | |
Online Access: | https://www.mdpi.com/1424-8220/22/6/2147 |
_version_ | 1797442453815427072 |
---|---|
author | Meng Wang Dazheng Feng Tingting Su Mohan Chen |
author_facet | Meng Wang Dazheng Feng Tingting Su Mohan Chen |
author_sort | Meng Wang |
collection | DOAJ |
description | Convolutional neural networks (CNNs) have significantly promoted the development of speaker verification (SV) systems because of their powerful deep feature learning capability. In CNN-based SV systems, utterance-level aggregation is an important component, and it compresses the frame-level features generated by the CNN frontend into an utterance-level representation. However, most of the existing aggregation methods aggregate the extracted features across time and cannot capture the speaker-dependent information contained in the frequency domain. To handle this problem, this paper proposes a novel attention-based frequency aggregation method, which focuses on the key frequency bands that provide more information for utterance-level representation. Meanwhile, two more effective temporal-frequency aggregation methods are proposed in combination with the existing temporal aggregation methods. The two proposed methods can capture the speaker-dependent information contained in both the time domain and frequency domain of frame-level features, thus improving the discriminability of speaker embedding. Besides, a powerful CNN-based SV system is developed and evaluated on the TIMIT and Voxceleb datasets. The experimental results indicate that the CNN-based SV system using the temporal-frequency aggregation method achieves a superior equal error rate of 5.96% on Voxceleb compared with the state-of-the-art baseline models. |
first_indexed | 2024-03-09T12:42:04Z |
format | Article |
id | doaj.art-a323c215a63341bcaaed092b59af8879 |
institution | Directory Open Access Journal |
issn | 1424-8220 |
language | English |
last_indexed | 2024-03-09T12:42:04Z |
publishDate | 2022-03-01 |
publisher | MDPI AG |
record_format | Article |
series | Sensors |
spelling | doaj.art-a323c215a63341bcaaed092b59af88792023-11-30T22:16:54ZengMDPI AGSensors1424-82202022-03-01226214710.3390/s22062147Attention-Based Temporal-Frequency Aggregation for Speaker VerificationMeng Wang0Dazheng Feng1Tingting Su2Mohan Chen3National Laboratory of Radar Signal Processing, Xidian University, Xi’an 710071, ChinaNational Laboratory of Radar Signal Processing, Xidian University, Xi’an 710071, ChinaNational Laboratory of Radar Signal Processing, Xidian University, Xi’an 710071, ChinaNational Laboratory of Radar Signal Processing, Xidian University, Xi’an 710071, ChinaConvolutional neural networks (CNNs) have significantly promoted the development of speaker verification (SV) systems because of their powerful deep feature learning capability. In CNN-based SV systems, utterance-level aggregation is an important component, and it compresses the frame-level features generated by the CNN frontend into an utterance-level representation. However, most of the existing aggregation methods aggregate the extracted features across time and cannot capture the speaker-dependent information contained in the frequency domain. To handle this problem, this paper proposes a novel attention-based frequency aggregation method, which focuses on the key frequency bands that provide more information for utterance-level representation. Meanwhile, two more effective temporal-frequency aggregation methods are proposed in combination with the existing temporal aggregation methods. The two proposed methods can capture the speaker-dependent information contained in both the time domain and frequency domain of frame-level features, thus improving the discriminability of speaker embedding. Besides, a powerful CNN-based SV system is developed and evaluated on the TIMIT and Voxceleb datasets. The experimental results indicate that the CNN-based SV system using the temporal-frequency aggregation method achieves a superior equal error rate of 5.96% on Voxceleb compared with the state-of-the-art baseline models.https://www.mdpi.com/1424-8220/22/6/2147convolutional neural networksspeaker verificationtemporal-frequency aggregationself-attention |
spellingShingle | Meng Wang Dazheng Feng Tingting Su Mohan Chen Attention-Based Temporal-Frequency Aggregation for Speaker Verification Sensors convolutional neural networks speaker verification temporal-frequency aggregation self-attention |
title | Attention-Based Temporal-Frequency Aggregation for Speaker Verification |
title_full | Attention-Based Temporal-Frequency Aggregation for Speaker Verification |
title_fullStr | Attention-Based Temporal-Frequency Aggregation for Speaker Verification |
title_full_unstemmed | Attention-Based Temporal-Frequency Aggregation for Speaker Verification |
title_short | Attention-Based Temporal-Frequency Aggregation for Speaker Verification |
title_sort | attention based temporal frequency aggregation for speaker verification |
topic | convolutional neural networks speaker verification temporal-frequency aggregation self-attention |
url | https://www.mdpi.com/1424-8220/22/6/2147 |
work_keys_str_mv | AT mengwang attentionbasedtemporalfrequencyaggregationforspeakerverification AT dazhengfeng attentionbasedtemporalfrequencyaggregationforspeakerverification AT tingtingsu attentionbasedtemporalfrequencyaggregationforspeakerverification AT mohanchen attentionbasedtemporalfrequencyaggregationforspeakerverification |