计算机工程与应用2017，Vol.53Issue(13)：49-54,6.DOI:10.3778/j.issn.1002-8331.1611-0217
基于深度自编码网络语音识别噪声鲁棒性研究
Research on noise robustness of speech recognition based on deep auto-encoder neural network
摘要
Abstract
To solve the problem of the center and the radius determined by randomly in the speech recognition tasks based on traditional Radial Basis Function(RBF)neural network, an unsupervised pre-training method which uses a large number of unlabeled data to initialize the network parameters is proposed to replace the traditional random initialization method based on the layered mechanism of human brain on speech recognition. This paper introduces the Deep Auto-Encoder(DAE) neural network as acoustical model and further analyzes robustness of speaker-independent isolated speech recognition on small size vocabulary database. The experimental results show that DAE outperforms RBF with MFCC(Mel Frequency Cepstrum Coefficient)feature extraction. In addition, compared to MFCC, GFCC(Gammatone Frequency Cepstrum Coefficient)gives more attribution on anti-noise property with a relative accuracy improvement of 1.87%in collaborate with DAE network.关键词
语音识别/鲁棒性/深度自编码网络/GFCC特征/MFCC特征Key words
speech recognition/robustness/Deep Auto-Encoder(DAE)neural network/Gammatone Frequency Ceps-trum Coefficient(GFCC)/Mel Frequency Cepstrum Coefficient(MFCC)分类
信息技术与安全科学引用本文复制引用
黄丽霞,王亚楠,张雪英,王洪翠..基于深度自编码网络语音识别噪声鲁棒性研究[J].计算机工程与应用,2017,53(13):49-54,6.基金项目
国家自然科学基金(No.61371193,No.61303109) （No.61371193,No.61303109）
山西省留学回国择优资助项目(晋人社厅函[2013]68号) （晋人社厅函[2013]68号）
山西省自然科学基金(No.2014021022-6). （No.2014021022-6）
