基于一阶有限差分商的带噪语音端点检测方法  

Speech Endpoint Detection Algorithm Based on First-order Finite Differential Quotient

作　　者：王振寰[1] 高炜[1,2] 梁立[1] 

机构地区：[1]云南师范大学信息学院,云南昆明650092 [2]苏州大学数学科学学院,江苏苏州215006

出　　处：《昆明学院学报》2011年第3期53-54,67,共3页Journal of Kunming University

基　　金：国家自然科学基金资助项目(60903131)

摘　　要：语音识别中的端点检测技术是语音识别的基础,低信噪比环境下的端点检测很大程度决定语音识别工作的有效进行.提出了一种利用一阶有限差分商结合折半查找来判定语音端点检测的方法,提高了端点检测的准确率.仿真实验表明,在噪声环境下该方法较好地实现了语音端点检测.Endpoint detection of speech recognition technology is the basis of voice recognition,endpoint detection with low SNR environment determined largely the work of speech recognition can be conducted effectively.This paper presents a method based on first-order finite differential quotient with a binary search to determine the endpoint detection;it improves the accuracy of endpoint detection.The emulate experiments show the proposed method can achieve better voice activity detection in a noisy environment.

关 键 词：语音识别 端点检测 特征提取 差分商 

分 类 号：TN912.3[电子电信—通信与信息系统]