The PhaseVocoder provides a basic phase vocoder that is initialized to N = 4096 and 75% overlap; the de-facto standard for PSM algorithms. The analysis and sythesis hop sizes are identical. More...

#include <fxobjects.h>

Public Member Functions
void	initialize (unsigned int _frameLength, unsigned int _hopSize, windowType _window)
	initialize the Fast FFT object for operation

void	destroyFFTW ()
	destroys the FFTW arrays and plans.

double	processAudioSample (double input, bool &fftReady)
	process one input sample throug the vocoder to produce one output sample

bool	addZeroPad (unsigned int count)
	zero pad the input timeline

bool	advanceAndCheckFFT ()
	advance the sample counter and check to see if we need to do the FFT.

fftw_complex *	getFFTData ()

fftw_complex *	getIFFTData ()

void	doInverseFFT ()
	perform the inverse FFT on the processed data

void	doOverlapAdd (double *outputData=nullptr, int length=0)
	perform the overlap/add on the IFFT data

unsigned int	getFrameLength ()

unsigned int	getHopSize ()

double	getOverlap ()

void	setOverlapAddOnly (bool b)

Protected Attributes
fftw_complex *	fft_input = nullptr
	array for FFT input

fftw_complex *	fft_result = nullptr
	array for FFT output

fftw_complex *	ifft_result = nullptr
	array for IFFT output

fftw_plan	plan_forward = nullptr
	FFTW plan for FFT.

fftw_plan	plan_backward = nullptr
	FFTW plan for IFFT.

double *	windowBuffer = nullptr
	array for window

double *	inputBuffer = nullptr
	input timeline (x)

double *	outputBuffer = nullptr
	output timeline (y)

unsigned int	inputWriteIndex = 0
	circular buffer index: input write

unsigned int	outputWriteIndex = 0
	circular buffer index: output write

unsigned int	inputReadIndex = 0
	circular buffer index: input read

unsigned int	outputReadIndex = 0
	circular buffer index: output read

unsigned int	wrapMask = 0
	input wrap mask

unsigned int	wrapMaskOut = 0
	output wrap mask

double	windowHopCorrection = 1.0
	window correction including hop/overlap

bool	needInverseFFT = false
	internal flag to signal IFFT required

bool	needOverlapAdd = false
	internal flag to signal overlap/add required

windowType	window = windowType::kHannWindow
	window type

unsigned int	frameLength = 0
	current FFT length

unsigned int	fftCounter = 0
	FFT sample counter.

unsigned int	hopSize = 0
	hop: ha = hs

double	overlap = 1.0
	overlap as raw value (75% = 0.75)

bool	overlapAddOnly = false
	flag for overlap-add-only algorithms

Detailed Description

The PhaseVocoder provides a basic phase vocoder that is initialized to N = 4096 and 75% overlap; the de-facto standard for PSM algorithms. The analysis and sythesis hop sizes are identical.

Audio I/O:

processes mono input into mono output.

Control I/F:

none.

Author: Will Pirkle http://www.willpirkle.com

Remarks: This object is included in Designing Audio Effects Plugins in C++ 2nd Ed. by Will Pirkle

Version: Revision : 1.0

Date: Date : 2018 / 09 / 7

Member Function Documentation

◆ addZeroPad()

bool PhaseVocoder::addZeroPad ( unsigned int count )

zero pad the input timeline

add zero-padding without advancing output read location, for fast convolution

NOTES:

Parameters

count the number of zero-valued samples to insert

Returns: true if the zero-insertion triggered a FFT event, false otherwise

◆ advanceAndCheckFFT()

bool PhaseVocoder::advanceAndCheckFFT ( )

advance the sample counter and check to see if we need to do the FFT.

increment the FFT counter and do the FFT if it is ready

NOTES:

Returns: true if the advancement triggered a FFT event, false otherwise

◆ destroyFFTW()

void PhaseVocoder::destroyFFTW ( )

destroys the FFTW arrays and plans.

destroy FFTW objects and plans

◆ doInverseFFT()

void PhaseVocoder::doInverseFFT ( )

perform the inverse FFT on the processed data

do the inverse FFT (optional; will be called automatically if not used)

NOTES:
This function is optional - if you need to sequence the output (synthesis) stage yourself
then you can call this function at the appropriate time - see the PSMVocoder object for an example

◆ doOverlapAdd()

void PhaseVocoder::doOverlapAdd	(	double *	outputData = `nullptr`,
		int	length = `0`
	)

perform the overlap/add on the IFFT data

do the overlap-add operation

NOTES:
This function is optional - if you need to sequence the output (synthesis) stage yourself
then you can call this function at the appropriate time - see the PSMVocoder object for an example

Parameters

outputData	an array of data to overlap/add: if this is NULL then the IFFT data is used
length	the lenght of the array of data to overlap/add: if this is -1, the normal IFFT length is used

◆ getFFTData()

fftw_complex * PhaseVocoder::getFFTData ( )

inline

get FFT data for manipulation (yes, naked pointer so you can manipulate)

◆ getFrameLength()

unsigned int PhaseVocoder::getFrameLength ( )

inline

get current FFT length

◆ getHopSize()

unsigned int PhaseVocoder::getHopSize ( )

inline

get current hop size ha = hs

◆ getIFFTData()

fftw_complex * PhaseVocoder::getIFFTData ( )

inline

get IFFT data for manipulation (yes, naked pointer so you can manipulate)

◆ getOverlap()

double PhaseVocoder::getOverlap ( )

inline

get current overlap as a raw value (75% = 0.75)

◆ initialize()

void PhaseVocoder::initialize	(	unsigned int	_frameLength,
		unsigned int	_hopSize,
		windowType	_window
	)

initialize the Fast FFT object for operation

setup the FFT for a given framelength and window type

NOTES:
See notes on symmetrical window arrays in comments.

Parameters

_frameLength	the FFT length - MUST be a power of 2
_hopSize	the hop size in samples: this object only supports ha = hs (pure real-time operation only)
_window	the window type (note: may be set to windowType::kNoWindow)

◆ processAudioSample()

double PhaseVocoder::processAudioSample	(	double	input,
		bool &	fftReady
	)

process one input sample throug the vocoder to produce one output sample

process audio sample through vocode; check fftReady flag to access FFT output

NOTES:

Parameters

input	the input sample x(n)
fftReady	a return flag indicating if the FFT has occurred and FFT data is ready to process

Returns: the vocoder output sample y(n)

◆ setOverlapAddOnly()

void PhaseVocoder::setOverlapAddOnly ( bool b )

inline

set the vocoder for overlap add only without hop-size

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

C:/RackAFX_v7/ASPIK_Doxygen/PluginObjects/fxobjects.h
C:/RackAFX_v7/ASPIK_Doxygen/PluginObjects/fxobjects.cpp

Public Member Functions

Protected Attributes

Detailed Description

Member Function Documentation

◆ addZeroPad()

◆ advanceAndCheckFFT()

◆ destroyFFTW()

◆ doInverseFFT()

◆ doOverlapAdd()

◆ getFFTData()

◆ getFrameLength()

◆ getHopSize()

◆ getIFFTData()

◆ getOverlap()

◆ initialize()

◆ processAudioSample()

◆ setOverlapAddOnly()