问题
I am using two methods,fft and lomb-scargle, to find periods for a timing sequence points.
However the fundamental frequency changes much even for the same dataset.
I can get the right result only when I fine tune the sampling rate.<>br/
How to choose the parameters of the two methods to get a consistent fundamental frequency.
How to know the result is reliable?
回答1:
The recent "Understanding the Lomb-Scargle Periodogram - Jacob T. VanderPlas" that you can find on This link to arxiv is a very good reference on how to chose the right set of parameters and how to test if your results are reliable.
回答2:
Fundamental pitch frequency can be very different from FFT magnitude peak. Pitch is a psycho-acoustic phenomena, and thus usually requires a different pitch detection/estimation method than just peak frequency estimation (Instead, try YAAPT, RAPT, cepstral/cepstrum, HPS, correlation lag estimators, & etc.)
来源：https://stackoverflow.com/questions/20782489/how-to-know-the-fundamental-frequency-is-right-detected-by-fft-lomb-scargle