给出了一种适合混响环境的语音建立信号ROTH加权时延估计方法。该方法首先通过房间混响模型和快速傅里叶变换提取语音建立信号;然后用语音建立信号来估计语音信号的互功率谱,并进行互功率谱的平滑处理;最后采用ROTH加权方法进行时延估计。仿真结果表明了该方法的有效性。
A ROTH weighted time delay estimation method is presented based on the speech onset signals in the reverberation environment. Firstly, the speech onset signals are obtained through the room reverberation model and the fast Fourier transform. Then, the speech onset signals are used to estimate the cross-power spectrum of the signals and smoothes it. Finally, the ROTH weighted method is used to estimate the time delay. The simulation result shows the effectiveness of the proposed method.