论文部分内容阅读
很多麦克风阵列时延估计算法在噪声和混响环境下性能都会下降。该文提出一种基于多路线性预测(multi-ple linear prediction,MLP)的时延估计算法。通过传递函数比估计来消除通道间传递函数的非对称性,提高信号相关程度;空间预测技术引入了阵列冗余信息,并以相关系数矩阵作为时延搜索的目标函数,提高时延估计的可靠性。实验结果显示了多路线性预测算法的估计准确率更高,性能更加稳健。与几种经典算法相比,在噪声和混响环境下MLP算法的估计正确率分别提高了5%和30%以上。
Many microphone array delay estimation algorithms experience degraded performance in both noisy and reverb environments. This paper presents a time-delay estimation algorithm based on multi-ple linear prediction (MLP). The transfer function ratio estimation is used to eliminate the asymmetry of the transfer function between channels and improve the signal correlation degree. Spatial prediction technology introduces array redundancy information and takes the correlation coefficient matrix as the objective function of time-delay search to improve the reliability of time-delay estimation Sex. The experimental results show that the estimation accuracy of multi-path linear prediction algorithm is higher and the performance is more robust. Compared with several classical algorithms, the estimated accuracy of MLP algorithm is improved by 5% and 30% in noise and reverberation respectively.