The harming part of room acoustics in automatic speech recognition
The harming part of room acoustics in automatic speech recognition
复制标题
自动语音识别中室内声学的危害部分
DOI：
10.21437/interspeech.2007-112
复制
发表时间：
2007
影响因子：
1.4
通讯作者：
R. Hoffmann
中科院分区：
文献类型：
--
作者：
Rico Petrick;Kevin Lohde;M. Wolff;R. Hoffmann
与训练条件相比，在真实室内场景中使用的自动语音识别（ASR）系统遭受不同的噪声和混响条件。本文介绍了一项旨在找出混响对语音识别危害最大的部分的研究。（cid:3）序列中的噪声被忽略。因此，在不同的房间和不同的扬声器到麦克风的距离测量和修正不同的真实房间脉冲响应（cid:2）。与相关的卷积脉冲响应的识别实验结果清楚地显示了早期和晚期以及高低频重复（cid:3）的依赖性。提出了一种脱噪方法的设计结论。
Automatic speech recognition (ASR) systems used in real indoor scenarios suffer from different noise and reverberation conditions compared to the training conditions. This article describes a study which aims to (cid:2)nd out what are the most harming parts of reverberation to speech recognition. Noise in(cid:3)uences are left out. Therefore different real room impulse responses in different rooms and different speaker to microphone distances are measured and modi(cid:2)ed. The results of the recognition experiments with the related convoluted impulse responses clearly show the dependency of early and late as well as high and low frequency re(cid:3)ections. Conclusions concerning the design of a dereverberation method are made.