Join GitHub today
much faster mpm auto-correlation #35
Today I met with a friend, Ron Schafer, co-author of the textbook "Discrete-Time Signal Processing" to get a better understanding of how the mpm auto-correlation works. While reviewing the code he figured out that it is possible to completely eliminate one of the three FFTs in acorr_r() and still get identical results in a much shorter time! Here's the new code, which I have validated. I suggest you try it and if you agree, replace your existing acorr_r code with this improvement.