Again returned back to What Every Programmer Should Know About Floating-Point Arithmetic. “LANGUAGE CHEAT SHEETS” section misses C/C++/Swift. Probably I should fix this. At least memset with 0 for floating point buffer bug should be reflected here. Nice that even [Apple employees make this errors(

Bookmarking IEEE-754 Analysis - nice tool for converting binaries/decimals/hex to 754 float/double.

I have to write my own example code like here about FFT. But without common errors initializing statically allocated buffers with memset (it’s actually double bug - not effective and memset of float buffer is not correct). Binary logarithm called natural. I have feelings that freq_bins is calculated incorrectly for common case. But I like approach, plots for every step, even hamming window, signal before/after window.