小空间占用的说话内容与说话者群同时识别  

Simultaneous recognition of both speech and speaker's group with small space occupation

作　　者：方晶[1,2,3] 朱嘉钢[1,2,3] 陆晓[3] 

机构地区：[1]江南大学物联网应用技术教育部工程研究中心,江苏无锡214122 [2]江南大学物联网工程学院,江苏无锡214122 [3]江南大学晓山股份联合实验室,江苏无锡214122

出　　处：《计算机应用研究》2015年第1期156-160,共5页Application Research of Computers

基　　金：江苏省产学研资助项目(BY2013015-40)

摘　　要：针对在嵌入式系统中说话内容与说话者群同时识别的应用环境,提出了小空间占用的说话内容与说话者群同时识别的方法。在近年关于说话内容与说话者身份同时识别机制的相关研究的基础上,用GMM替换HMM对说话内容与说话者群语音特征进行建模与投票机制完成说话者群识别以降低算法的空间占用,并用SQ(soft quantization)对多个语音内容识别器的识别结果集成输出。当集成模型为六个时,减少20.88%语音内容识别错误率,说话者群平均识别率达到81.57%。实验结果证实了所提算法的可行性。In order to meet the requirement environment in which both speech and speaker＇s group were simultaneously recog- nized through embedded system, this paper proposed a new method with low space consumption for simultaneous recognition of both speech and speaker＇s group. On the basis of research on mechanism of speech recognition and speaker identification in recent years, GMM instead of HMM, was used to model the voice characteristics of speech and speaker and voting mechanism to complete the speaker＇s group identification to reduce space occupation. SQ（ soft quantization） was used to integrate the out- puts from multiple speech recognizer. When integrating 6 models, it reduced 20.88% error rate in Speech recognition and achieved 81.57% average recognition rate in speaker＇s group. The experimental results confirm feasibility of the proposed algorithm,

关 键 词：语音识别 说话者群识别 集成学习 GMM 嵌入式系统 投票机制 SOFT QUANTIZATION 

分 类 号：TP391.42[自动化与计算机技术—计算机应用技术]