语音信号频谱的获取  

Obtainment of Speech Spectrum

作　　者：罗海涛[1] LUO Hai-tao(Guangdong University of Foreign Studies,Guangzhou 510420,Guangdong)

机构地区：[1]广东外语外贸大学信息学院,广东广州510420

出　　处：《电脑与电信》2022年第4期55-57,共3页Computer & Telecommunication

摘　　要：语言是人类特有的能力。通过声音传感器,把语音转换为电信号;通过采样和幅度量化,把语音信号转化为一系列离散化的采样数据,便于计算机存储、传输和处理;通过快速傅里叶变换,可以得到采样数据的频谱。频域特征与时域特征结合可应用于语音的识别,说话人识别,语音的编码、分割、合成、端点检测等领域。Speech is an unique ability of human beings.Speech can be transformed to electric signal through speech sensor,and be transformed to sampling data further through sampling.Speech characteristics in frequency domain include spectrum which can be obtained through Fast Fourier Transform(FFT).Speech characteristics in time and frequency domain can be applied in speech recognition,speaker recognition,speech coding,and voice activity detection(VAD).

关 键 词：语音信号 采样数据 分帧 快速傅里叶变换 频谱 

分 类 号：TN912.3[电子电信—通信与信息系统]