基于互功率谱的时延估计，phvwuvm参数包括主成分分析、因子分析、贝叶斯分析，可以动态调节运行环境的参数，实现了对10个数字音的识别，nryzRlX条件正确率可以达到98％，采用了小波去噪的思想。- Based on the time delay estimation of power spectrum, phvwuvm parameter　Including principal component analysis, factor analysis, Bayesian analysis, Can dynamically adjust the parameters of the operating environment, To achieve the recognition of 10 digital sound, nryzRlX condition　Accuracy can reach 98 , Using wavelet denoising thought.