An Evaluation of Word-Level Confidence Estimation for End-to-End Automatic Speech Recognition

Dan Oneaţă, Alexandru Caranica, Adriana Stan, H. Cucu
{"title":"An Evaluation of Word-Level Confidence Estimation for End-to-End Automatic Speech Recognition","authors":"Dan Oneaţă, Alexandru Caranica, Adriana Stan, H. Cucu","doi":"10.1109/SLT48900.2021.9383570","DOIUrl":null,"url":null,"abstract":"Quantifying the confidence (or conversely the uncertainty) of a prediction is a highly desirable trait of an automatic system, as it improves the robustness and usefulness in downstream tasks. In this paper we investigate confidence estimation for end-to-end automatic speech recognition (ASR). Previous work has addressed confidence measures for lattice-based ASR, while current machine learning research mostly focuses on confidence measures for unstructured deep learning. However, as the ASR systems are increasingly being built upon deep end-to-end methods, there is little work that tries to develop confidence measures in this context. We fill this gap by providing an extensive benchmark of popular confidence methods on four well-known speech datasets. There are two challenges we overcome in adapting existing methods: working on structured data (sequences) and obtaining confidences at a coarser level than the predictions (words instead of tokens). Our results suggest that a strong baseline can be obtained by scaling the logits by a learnt temperature, followed by estimating the confidence as the negative entropy of the predictive distribution and, finally, sum pooling to aggregate at word level.","PeriodicalId":243211,"journal":{"name":"2021 IEEE Spoken Language Technology Workshop (SLT)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2021-01-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"17","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2021 IEEE Spoken Language Technology Workshop (SLT)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/SLT48900.2021.9383570","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 17

Abstract

Quantifying the confidence (or conversely the uncertainty) of a prediction is a highly desirable trait of an automatic system, as it improves the robustness and usefulness in downstream tasks. In this paper we investigate confidence estimation for end-to-end automatic speech recognition (ASR). Previous work has addressed confidence measures for lattice-based ASR, while current machine learning research mostly focuses on confidence measures for unstructured deep learning. However, as the ASR systems are increasingly being built upon deep end-to-end methods, there is little work that tries to develop confidence measures in this context. We fill this gap by providing an extensive benchmark of popular confidence methods on four well-known speech datasets. There are two challenges we overcome in adapting existing methods: working on structured data (sequences) and obtaining confidences at a coarser level than the predictions (words instead of tokens). Our results suggest that a strong baseline can be obtained by scaling the logits by a learnt temperature, followed by estimating the confidence as the negative entropy of the predictive distribution and, finally, sum pooling to aggregate at word level.
端到端自动语音识别中词级置信度估计的评价
量化预测的置信度(或相反的不确定性)是自动化系统非常需要的特性,因为它提高了下游任务的鲁棒性和有用性。本文研究了端到端自动语音识别(ASR)的置信度估计。以前的工作已经解决了基于网格的ASR的置信度度量,而当前的机器学习研究主要集中在非结构化深度学习的置信度度量上。然而,随着ASR系统越来越多地建立在深度端到端方法之上,很少有工作试图在这种情况下开发信心措施。我们通过在四个知名语音数据集上提供广泛的流行置信度方法基准来填补这一空白。在调整现有方法时,我们克服了两个挑战:处理结构化数据(序列),并在比预测(单词而不是令牌)更粗糙的级别上获得置信度。我们的结果表明,可以通过通过学习温度缩放logits来获得强基线,然后将置信度估计为预测分布的负熵,最后在单词水平上进行和池汇总。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
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