摘要
针对说话人识别性能在噪声环境下急剧下降的问题,提出了基于调制域谱减法的鲁棒性说话人识别方法。首先在说话人识别前端通过调制域谱减法对含噪语音进行增强处理,然后通过Gammatone滤波器组提取对噪声具有抑制作用的特征,最后与说话人模型进行匹配识别。仿真结果表明,运用此方法能显著抑制噪声对说话人识别系统的影响,提高系统的识别率。
In order to solve the problem that recognition rate decreases significantly under noisy environment,a robust speaker recognition method based on modulation domain spectral subtraction is proposed. The method uses spectral subtraction in modulation domain to improve the intelligibility at the front-end,then extracts the robust speaker feature through Gammatone filter bank,and finally the feature will be matched with the model of the speaker. The simulation results show that the proposed method can significantly suppress the effect of noise on the speaker recognition system and thus improve the recognition rate of the system.
作者
程小伟
王健
曾庆宁
谢先明
龙超
CHENG Xiao -wei WANG Jian ZENG Qing - ning XIE Xian-ming LONG Chao(School of Information and Communication, Guilin University of Electronic Technology, Guilin 541004, P. R. China)
出处
《科学技术与工程》
北大核心
2017年第3期252-257,共6页
Science Technology and Engineering
基金
国家自然科学基金(61461011)
广西自然科学基金(2014GXNSFBA118273)资助