Improved arithmetic of wavelet transformation for pitch frequency detection
一种改进的小波变换基音检测算法

In order to overcome the disadvantages of low accuracy, high complexity and lack of robustness in many actual pitch frequency algorithms, this paper proposes the method that uses wavelet transformation with linear forecast for removing noise and the resonant influence. And an improved autocorrelation function and average range difference function method is used to test the peak position. The simulation experiment shows that the approach is superior to the method of autocorrelation function and the method of average magnitude difference function. Not only does the method reduce the frequency and half frequency of times error, improving the frequency to extract the extraction of pitch precision, but also is under the lower circumstance signal-to-noise ratio. It can accurately extract pitch frequency.