Whisper-SV:为低数据资源扬声器验证调整 Whisper

Li Zhang, Ning Jiang, Qing Wang, Yue Li, Quan Lu, Lei Xie
{"title":"Whisper-SV:为低数据资源扬声器验证调整 Whisper","authors":"Li Zhang, Ning Jiang, Qing Wang, Yue Li, Quan Lu, Lei Xie","doi":"arxiv-2407.10048","DOIUrl":null,"url":null,"abstract":"Trained on 680,000 hours of massive speech data, Whisper is a multitasking,\nmultilingual speech foundation model demonstrating superior performance in\nautomatic speech recognition, translation, and language identification.\nHowever, its applicability in speaker verification (SV) tasks remains\nunexplored, particularly in low-data-resource scenarios where labeled speaker\ndata in specific domains are limited. To fill this gap, we propose a\nlightweight adaptor framework to boost SV with Whisper, namely Whisper-SV.\nGiven that Whisper is not specifically optimized for SV tasks, we introduce a\nrepresentation selection module to quantify the speaker-specific\ncharacteristics contained in each layer of Whisper and select the top-k layers\nwith prominent discriminative speaker features. To aggregate pivotal\nspeaker-related features while diminishing non-speaker redundancies across the\nselected top-k distinct layers of Whisper, we design a multi-layer aggregation\nmodule in Whisper-SV to integrate multi-layer representations into a singular,\ncompacted representation for SV. In the multi-layer aggregation module, we\nemploy convolutional layers with shortcut connections among different layers to\nrefine speaker characteristics derived from multi-layer representations from\nWhisper. In addition, an attention aggregation layer is used to reduce\nnon-speaker interference and amplify speaker-specific cues for SV tasks.\nFinally, a simple classification module is used for speaker classification.\nExperiments on VoxCeleb1, FFSVC, and IMSV datasets demonstrate that Whisper-SV\nachieves EER/minDCF of 2.22%/0.307, 6.14%/0.488, and 7.50%/0.582, respectively,\nshowing superior performance in low-data-resource SV scenarios.","PeriodicalId":501178,"journal":{"name":"arXiv - CS - Sound","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Whisper-SV: Adapting Whisper for Low-data-resource Speaker Verification\",\"authors\":\"Li Zhang, Ning Jiang, Qing Wang, Yue Li, Quan Lu, Lei Xie\",\"doi\":\"arxiv-2407.10048\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Trained on 680,000 hours of massive speech data, Whisper is a multitasking,\\nmultilingual speech foundation model demonstrating superior performance in\\nautomatic speech recognition, translation, and language identification.\\nHowever, its applicability in speaker verification (SV) tasks remains\\nunexplored, particularly in low-data-resource scenarios where labeled speaker\\ndata in specific domains are limited. To fill this gap, we propose a\\nlightweight adaptor framework to boost SV with Whisper, namely Whisper-SV.\\nGiven that Whisper is not specifically optimized for SV tasks, we introduce a\\nrepresentation selection module to quantify the speaker-specific\\ncharacteristics contained in each layer of Whisper and select the top-k layers\\nwith prominent discriminative speaker features. To aggregate pivotal\\nspeaker-related features while diminishing non-speaker redundancies across the\\nselected top-k distinct layers of Whisper, we design a multi-layer aggregation\\nmodule in Whisper-SV to integrate multi-layer representations into a singular,\\ncompacted representation for SV. In the multi-layer aggregation module, we\\nemploy convolutional layers with shortcut connections among different layers to\\nrefine speaker characteristics derived from multi-layer representations from\\nWhisper. In addition, an attention aggregation layer is used to reduce\\nnon-speaker interference and amplify speaker-specific cues for SV tasks.\\nFinally, a simple classification module is used for speaker classification.\\nExperiments on VoxCeleb1, FFSVC, and IMSV datasets demonstrate that Whisper-SV\\nachieves EER/minDCF of 2.22%/0.307, 6.14%/0.488, and 7.50%/0.582, respectively,\\nshowing superior performance in low-data-resource SV scenarios.\",\"PeriodicalId\":501178,\"journal\":{\"name\":\"arXiv - CS - Sound\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-07-14\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - CS - Sound\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2407.10048\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - CS - Sound","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2407.10048","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

摘要

经过 68 万小时海量语音数据的训练,Whisper 是一种多任务、多语言语音基础模型,在自动语音识别、翻译和语言识别方面表现出卓越的性能。然而,它在说话人验证(SV)任务中的适用性仍有待探索,尤其是在低数据资源场景中,特定领域中的标注说话人数据有限。为了填补这一空白,我们提出了一个轻量级适配器框架,即 Whisper-SV,以利用 Whisper 提升 SV。鉴于 Whisper 并未专门针对 SV 任务进行优化,我们引入了呈现选择模块,以量化 Whisper 每一层所包含的特定说话人特征,并选择具有突出辨别说话人特征的 top-k 层。为了聚合与说话人相关的关键特征,同时减少 Whisper 中选出的前 k 个不同层中的非说话人冗余,我们在 Whisper-SV 中设计了一个多层聚合模块,将多层表征整合为一个单一、紧凑的 SV 表征。在多层聚合模块中,我们利用卷积层与不同层之间的快捷连接来提炼从 Whisper 的多层表征中得出的说话者特征。在 VoxCeleb1、FFSVC 和 IMSV 数据集上的实验表明,Whisper-SV 的 EER/minDCF 分别为 2.22%/0.307、6.14%/0.488 和 7.50%/0.582,在低数据资源 SV 场景中表现出了卓越的性能。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
Whisper-SV: Adapting Whisper for Low-data-resource Speaker Verification
Trained on 680,000 hours of massive speech data, Whisper is a multitasking, multilingual speech foundation model demonstrating superior performance in automatic speech recognition, translation, and language identification. However, its applicability in speaker verification (SV) tasks remains unexplored, particularly in low-data-resource scenarios where labeled speaker data in specific domains are limited. To fill this gap, we propose a lightweight adaptor framework to boost SV with Whisper, namely Whisper-SV. Given that Whisper is not specifically optimized for SV tasks, we introduce a representation selection module to quantify the speaker-specific characteristics contained in each layer of Whisper and select the top-k layers with prominent discriminative speaker features. To aggregate pivotal speaker-related features while diminishing non-speaker redundancies across the selected top-k distinct layers of Whisper, we design a multi-layer aggregation module in Whisper-SV to integrate multi-layer representations into a singular, compacted representation for SV. In the multi-layer aggregation module, we employ convolutional layers with shortcut connections among different layers to refine speaker characteristics derived from multi-layer representations from Whisper. In addition, an attention aggregation layer is used to reduce non-speaker interference and amplify speaker-specific cues for SV tasks. Finally, a simple classification module is used for speaker classification. Experiments on VoxCeleb1, FFSVC, and IMSV datasets demonstrate that Whisper-SV achieves EER/minDCF of 2.22%/0.307, 6.14%/0.488, and 7.50%/0.582, respectively, showing superior performance in low-data-resource SV scenarios.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信