TD_LTE移动终端NC系统VAD模块的改进和实现

经过详细研究信号和噪声时频域的不同特性，提出了一种改进的VAD算法。改进前的VAD算法是一种静态的语音停顿周期检测算法，在信噪比很低时，检测性能变差；改进后的VAD算法在信号的全频带，低频带和高频带动态跟踪信号的短时功率包络，多次门限比较后，当出现语音停顿周期时，将做出基于帧的判决。为了验证改进VAD算法在TD_LTE手机NC系统中的应用能力，测试了在不同噪声、不同信噪比情况下的NC系统指标，并与传统NC系统做了比较。经评测，含新VAD算法的改进后NC算法能提高信噪比，并帮助TD_LTE移动终端顺利通过入网测试。

Implementation of VAD Module in NCSystemof TD_LTE Mobile Terminal

Considered the time and frequency domain different characteristics of sigalal and noise, a improved algortthm of the V AD fit for the downlink NC system is proposed. The classic VAD algorithm is a static detection algorithms for the voice actively detected. The detection performance has deteriorated while the SNR is low. The improved VAD algorithms has dynamic tracked the signals envelope in all band, low-fre- quency band and high-frequency hand. It has decided whether or not voice after more the threshold compared. The academic evaluations and practice tests proved that the algorithm can more precisely detect the noise tone of the NC system with new VAD method, helping the TD_LTE mobile terminal passing the network test.