一种用于3G系统中复杂背景噪声环境下的话音激活检测算法

A novel VAD algorithm applied for 3G system in complex background noise environment

本文讨论了一种新的应用在 3G自适应多速率系统中复杂背景噪声环境下的话音激活检测算法。这种算法基于谱估计理论和周期信号检测方法 ,应用一个IIR滤波器组把输入窄带语音信号分成九个频带 ,进而估计每个频带的语音信号和背景噪声的电平 ,结合基音和音调检测 ,在区分语音和移动环境中的大多数背景噪声时表现得足够健壮。最后 ,基于欧洲电信标准委员会建议的 3G平台对这种算法进行了仿真评估 。

A novel voice activity detection(VAD)algorithm applied for adaptive multi rate(AMR)CODEC system is discussed in this paper The algorithm is based on spectral estimation and periodical detection,which uses an IIR filter bank to divide the input speech signal into nine bands from 0 to 4000Hz Moreover,the levels of speech and background noise are estimated,with the result of pitch and tone detection,the algorithm is robust enough to most kinds of background noise Finally,the algorithm is simulated in the 3G system platform recommended by ETSI and analyzed in terms of subjective and objective method supported by ITU T