基于连续小波变换(CWT)的汉语语音谱图

The Chinese Spectrograms Based on Continuous Wavelet Transform (CWT)

本文从理论上分析了小波变换应用于语音分析的优势所在，在将小波变换直接应用于语音分析的基础上，结合汉语语音信号的特点，对小波变换的相平面作了调整，并将其应用于语音的谱分析中。结果证明，调整后的相平面，更加适合于汉语语音信号的分析，可以更清晰的表达语音信号的动态谱结构。

In this paper, we first analyses the advantage of using wavelet transform in speech analysis in theory. After applying wavelet transform to sped analysis directly, we reform the phase plane of WT for the sake of chinese speech characteristic and then applying the new phase plane to spectrum analysis. The experiments shown that the new phase plane is more effective in analyzing the signal of Chinese speech. It can describe the dynamic spectrum constitution more clearly.