Driver for the fixedpoint-fft library (https://github.com/pschatzmann/fixedpoint-fft). CalcT selects the calculation type used by the underlying engine: int8_t/int16_t/int32_t for fixed point Qn math (smallest/fastest, ideal for plain AVR), or float/double on FPU-equipped targets (ESP32, Cortex-M4F/M7, desktop). Defaults to int16_t (Q15), which gives good precision at any FFT size without needing an FPU. The FFTDriver interface itself is always float, so values are rescaled into/out of CalcT's Qn range on the way in/out via fixedpoint_fft::fftConvertSample. More...

#include <AudioFixedFFT.h>

Inheritance diagram for FFTDriverFixedFFT< CalcT >:

Public Member Functions
bool	begin (int len) override

void	end () override

fixedpoint_fft::FixedFFTEngineCore< CalcT > *	engine ()
	Provides access to the underlying fixedpoint-fft engine.

void	fft () override
	Perform FFT.

bool	getBin (int pos, FFTBin &bin) override
	gets the value of a bin

float	getValue (int idx) override
	get Real value

bool	isReverseFFT () override
	Returns true if reverse FFT is supported.

bool	isValid () override

float	magnitude (int idx) override
	Calculate the magnitude (fft result) at index (sqr(i² + r²))

float	magnitudeFast (int idx) override
	magnitude w/o sqrt

void	rfft () override
	Inverse fft - convert fft result back to time domain (samples)

bool	setBin (int pos, FFTBin &bin)
	sets the value of a bin

bool	setBin (int pos, float real, float img) override
	sets the value of a bin

void	setValue (int idx, float value) override
	Sets the real value.

Protected Member Functions
float	imagAsFloat (int idx)

float	realAsFloat (int idx)

Protected Attributes
fixedpoint_fft::FixedFFTEngineCore< CalcT >	ffp_object

int	len = 0

bool	valid = false

Detailed Description

template<typename CalcT = int16_t>
class audio_tools::FFTDriverFixedFFT< CalcT >

Driver for the fixedpoint-fft library (https://github.com/pschatzmann/fixedpoint-fft). CalcT selects the calculation type used by the underlying engine: int8_t/int16_t/int32_t for fixed point Qn math (smallest/fastest, ideal for plain AVR), or float/double on FPU-equipped targets (ESP32, Cortex-M4F/M7, desktop). Defaults to int16_t (Q15), which gives good precision at any FFT size without needing an FPU. The FFTDriver interface itself is always float, so values are rescaled into/out of CalcT's Qn range on the way in/out via fixedpoint_fft::fftConvertSample.

Author: Phil Schatzmann

Copyright: GPLv3

Member Function Documentation

◆ begin()

template<typename CalcT = int16_t>

bool begin ( int len )

inlineoverridevirtual

Implements FFTDriver.

◆ end()

template<typename CalcT = int16_t>

void end ( )

inlineoverridevirtual

Implements FFTDriver.

◆ engine()

template<typename CalcT = int16_t>

fixedpoint_fft::FixedFFTEngineCore< CalcT > * engine ( )

inline

Provides access to the underlying fixedpoint-fft engine.

◆ fft()

template<typename CalcT = int16_t>

void fft ( )

inlineoverridevirtual

Perform FFT.

Implements FFTDriver.

◆ getBin()

template<typename CalcT = int16_t>

bool getBin	(	int	pos,
		FFTBin &	bin
	)

inlineoverridevirtual

gets the value of a bin

Reimplemented from FFTDriver.

◆ getValue()

template<typename CalcT = int16_t>

float getValue ( int idx )

inlineoverridevirtual

get Real value

Implements FFTDriver.

◆ imagAsFloat()

template<typename CalcT = int16_t>

float imagAsFloat ( int idx )

inlineprotected

◆ isReverseFFT()

template<typename CalcT = int16_t>

bool isReverseFFT ( )

inlineoverridevirtual

Returns true if reverse FFT is supported.

Reimplemented from FFTDriver.

◆ isValid()

template<typename CalcT = int16_t>

bool isValid ( )

inlineoverridevirtual

Implements FFTDriver.

◆ magnitude()

template<typename CalcT = int16_t>

float magnitude ( int idx )

inlineoverridevirtual

Calculate the magnitude (fft result) at index (sqr(i² + r²))

Implements FFTDriver.

◆ magnitudeFast()

template<typename CalcT = int16_t>

float magnitudeFast ( int idx )

inlineoverridevirtual

magnitude w/o sqrt

Implements FFTDriver.

◆ realAsFloat()

template<typename CalcT = int16_t>

float realAsFloat ( int idx )

inlineprotected

◆ rfft()

template<typename CalcT = int16_t>

void rfft ( )

inlineoverridevirtual

Inverse fft - convert fft result back to time domain (samples)

Reimplemented from FFTDriver.

◆ setBin() [1/2]

bool setBin	(	int	pos,
		FFTBin &	bin
	)

inlineinherited

sets the value of a bin

◆ setBin() [2/2]

template<typename CalcT = int16_t>

bool setBin	(	int	idx,
		float	real,
		float	img
	)

inlineoverridevirtual

sets the value of a bin

Reimplemented from FFTDriver.

◆ setValue()

template<typename CalcT = int16_t>

void setValue	(	int	pos,
		float	value
	)

inlineoverridevirtual

Sets the real value.

Implements FFTDriver.

Member Data Documentation

◆ ffp_object

template<typename CalcT = int16_t>

fixedpoint_fft::FixedFFTEngineCore<CalcT> ffp_object

protected

◆ len

template<typename CalcT = int16_t>

int len = 0

protected

◆ valid

template<typename CalcT = int16_t>

bool valid = false

protected

The documentation for this class was generated from the following file:

src/AudioTools/FFT/AudioFixedFFT.h

Public Member Functions

Protected Member Functions

Protected Attributes

Detailed Description

Member Function Documentation

◆ begin()

◆ end()

◆ engine()

◆ fft()

◆ getBin()

◆ getValue()

◆ imagAsFloat()

◆ isReverseFFT()

◆ isValid()

◆ magnitude()

◆ magnitudeFast()

◆ realAsFloat()

◆ rfft()

◆ setBin() [1/2]

◆ setBin() [2/2]

◆ setValue()

Member Data Documentation

◆ ffp_object

◆ len

◆ valid