基于声纹识别的智能照明语音识别算法研究  

Research on intelligent lighting speech recognition algorithm based on voiceprint recognition

作　　者：王建平[1] 马兰兰 孙伟[1] WANG Jianping;MA Lanlan;SUN Wei(School of Electrical Engineering and Automation,Hefei University of Technology,Hefei 230009,China)

机构地区：[1]合肥工业大学电气与自动化工程学院,安徽合肥230009

出　　处：《传感器与微系统》2020年第6期37-40,44,共5页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(51877060)。

摘　　要：提出了一种基于声纹识别的智能照明语音识别算法。采用短时能量和短时平均过零率判别声控信号;采用基于高斯混合模型-通用背景模型(GMM-UBM)的方法判定声控人的身份;采用基于动态时间规整(DTW)的方法识别语义。通过建立多个身份人和多种声控命令的自适应实时语音训练样本库,采用对数似然分法和矢量累积距离法实现声控人身份认定与声控信号语义识别。仿真实验结果表明:该算法能快速准确判定与识别智能照明声控信号的身份和语义。An intelligent illumination speech recognition algorithm based on voiceprint recognition is proposed.The short-time energy and short-term average zero-crossing rate are used to judge the voice signal. The method based on Gaussian mixture model-universal background model( GMM-UBM) is used to determine the identity of the controller. The method based on the dynamic time warping( DTW) is used to identify semantics. Through the establishment of adaptive real-time speech training sample library with multiple identity persons and multiple voice commands,the logarithm likelihood method and vector cumulative distance method are used to realize voice controller identification and voice signal semantic recognition. The simulation results show that the algorithm can quickly and accurately determine and identify the identity and semantics of intelligent lighting voice signal.

关 键 词：声纹识别 智能照明 高斯混合模型-通用背景模型 动态时间规整 

分 类 号：TN912.34[电子电信—通信与信息系统]