摘要
语音端点检测直接决定了语音识别的精度和速度。车载环境是一个非常复杂的环境,信噪比(SNR)有可能出现很低的情况,对于传统的时域端点检测方法来说,在这种环境下的端点检测效果很差,而双门限在高信噪比条件下,端点检测的效果非常好,识别率很高,这就使得提高车载环境下语音SNR非常关键。文章提出采用改进的小波去噪和改进的双门限方法进行端点检测。实验结果表明,综合改进小波去噪和改进双门限的方法虽然有一定量的信号失真,但失真在可接受范围之内,并且在不增大运算量的情况下端点检测的效果比传统的双门限效果要好,表明了本文算法的有效性。
The endpoint detection is an important part in signal processing. Endpoint detection directly determines the accuracy and speed of the voice recognition. Car environment is a very complex environment,the signal-to-noise ratio of the signal possibility is very low,for the traditional time domain endpoint detection method,in this environment the endpoint detection effect is very poor. The double door limit under the condition of high SNR,endpoint detection effect is very good,the recognition rate is very high,this makes the prompt on-board environment voice SNR is critical. In this paper,the improved wavelet denoising and the improved double threshold algorithm is adopted for endpoint detection. The experimental results show that the integrated method of wavelet denoising and double threshold though there is a certain amount of signal distortion,the distortion in the range of acceptable,and in the case of not increase the computational complexity,the endpoint detection effect is better than traditional double threshold effect,which show the effectiveness of the algorithm in this paper.
作者
张恒
周萍
Zhang Heng Zhou Ping(Shcool of Electronic Engineering and Automation, Guilin University of Electronic Technology, Guilin 541004, China)
出处
《微型机与应用》
2017年第5期21-23,共3页
Microcomputer & Its Applications
基金
广西研究生教育创新计划资助项目(YCSZ2015152)
关键词
车载环境
小波去噪
双门限
端点检测
on-board environment
wavelet denoising
dual-threshold
endpoint dectect
