signalsmith::fft::FFT< V > Class Template Reference

#include <fft.h>

Detailed Description

template<typename V = double>
class signalsmith::fft::FFT< V >

Floating-point FFT implementation.

It is fast for 2^a * 3^b. Here are the peak and RMS errors for float/double computation:

Simulated errors for pure-tone harmonic inputs, compared to a theoretical upper bound from "Roundoff error analysis of the fast Fourier transform" (G. Ramos 1971)

Static Methods
static size_t	fastSizeAbove (size_t size)
static size_t	fastSizeBelow (size_t size)

Methods
	FFT (size_t size, int fastDirection=0)
size_t	setSize (size_t size)
size_t	setFastSizeAbove (size_t size)
size_t	setFastSizeBelow (size_t size)
const size_t &	size () const
template<typename InputIterator, typename OutputIterator>
void	fft (InputIterator &&input, OutputIterator &&output)
template<typename InputIterator, typename OutputIterator>
void	ifft (InputIterator &&input, OutputIterator &&output)