3G系统中复杂背景噪声环境下话音激活检测算法性能分析

Performance Analysis of VAD Algorithm Applied for 3G System in Complex Background Noise Environment

以第三代移动通信合作项目组 ( 3GPP)和欧洲电信标准委员会 ( ETSI) GSM相关标准为依据 ,分析自适应多速率语音编码器中话音激活检测算法在不同背景噪声和不同输入信噪比下的性能 .基于对大量实验结果的分析 ,对其在 3G系统中应用的鲁棒性加以验证 ,同时和国际电联电信标准组 ( ITU- T)建议的 G.72 9算法进行比较 .研究表明 ,该算法在非平稳背景噪声环境下的性能优于 G.72 9,在保证重构语音足够高可懂度和自然度的前提下 ,使整个移动通信系统容量增加约 30 % .

In terms of standards recommended by 3GPP and ETSI, the performance of voice activity detection (VAD) algorithm applied for adaptive multi rate(AMR) speech codec was simulated and analyzed with varying background noise and different signal to noise ratios. Based on a great deal of test results, the robustness of AMR VAD applied for 3G system was verified and compared with ITU G.729 VAD. It proves that the AMR VAD performs better than G.729 VAD in nonstationary background noise environment, the capacity of mobile communication system is expanded about 30% when the intelligibility and naturalness of reconstructed speech is not impaired.