arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2008.06580eess.AScs.CLcs.SD
收藏

基于神经网络的语音识别自适应算法综述

Adaptation Algorithms for Neural Network-Based Speech Recognition: An Overview

University of Edinburgh(爱丁堡大学);JPMorgan Chase(摩根大通);Microsoft Corporation(微软公司);University of New South Wales(新南威尔士大学);Apple(苹果公司)

Peter Bell, Joachim Fainberg, Ondrej Klejch, Jinyu Li, Steve Renals, Pawel Swietojanski

摘要:

我们提出了一个结构化的基于神经网络的语音识别的自适应算法的概述,考虑混合隐马尔可夫模型/神经网络系统和端到端的神经网络系统,重点是说话人自适应,域自适应,和口音自适应。概述的特点自适应算法的基础上嵌入,模型参数自适应,或数据增强。我们提出了一个荟萃分析的性能语音识别自适应算法,根据文献中报道的相对错误率降低。

英文摘要:

We present a structured overview of adaptation algorithms for neural network-based speech recognition, considering both hybrid hidden Markov model / neural network systems and end-to-end neural network systems, with a focus on speaker adaptation, domain adaptation, and accent adaptation. The overview characterizes adaptation algorithms as based on embeddings, model parameter adaptation, or data augmentation. We present a meta-analysis of the performance of speech recognition adaptation algorithms, based on relative error rate reductions as reported in the literature.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑