Audio::FUniformPartitionConvolution Class Reference

#include <UniformPartitionConvolution.h>

Inheritance diagram for Audio::FUniformPartitionConvolution:

Public Types
using	FSharedFFTRef = TSharedRef< IFFTAlgorithm >

Public Member Functions
	FUniformPartitionConvolution (const FUniformPartitionConvolutionSettings &InSettings, FSharedFFTRef InFFTAlgorithm)
virtual	~FUniformPartitionConvolution ()
virtual int32	GetNumSamplesInBlock () const override
virtual int32	GetNumAudioInputs () const override
virtual int32	GetNumAudioOutputs () const override
virtual void	ProcessAudioBlock (const float *const InSamples[], float *const OutSamples[]) override
virtual void	ResetAudioHistory () override
virtual int32	GetMaxNumImpulseResponseSamples () const override
virtual int32	GetNumImpulseResponses () const override
virtual int32	GetNumImpulseResponseSamples (int32 InImpulseResponseIndex) const override
virtual void	SetImpulseResponse (int32 InImpulseResponseIndex, const float *InSamples, int32 InNumSamples) override
virtual void	SetMatrixGain (int32 InAudioInputIndex, int32 InImpulseResponseIndex, int32 InAudioOutputIndex, float InGain) override
virtual float	GetMatrixGain (int32 InAudioInputIndex, int32 InImpulseResponseIndex, int32 InAudioOutputIndex) const override
Public Member Functions inherited from Audio::IConvolutionAlgorithm
virtual	~IConvolutionAlgorithm ()

Detailed Description

FUniformPartitionConvolution

FUniformPartitionConvolution implements a fft based convolution algorithm supporting multiple inputs, impulse responses and outputs. The algorithm subdivides an impulse response into uniform blocks which results in several positive side effects:

1. Latency is set to the block size which in many cases is less than the length of the impulse response.
2. Increased performance over brute time domain convolution.
3. Reuse of fourier transformed data when inputs are mapped to multiple outputs.

The block size and latency is determined by the FFT algorithm passed to the constructor. BlockSize = FFTAlgorithm->Size() / 2 Latency = FFTAlgorithm->Size() / 2

Member Typedef Documentation

FSharedFFTRef

using Audio::FUniformPartitionConvolution::FSharedFFTRef = TSharedRef<IFFTAlgorithm>

Shared reference to FFT algorithm.

Constructor & Destructor Documentation

FUniformPartitionConvolution()

Audio::FUniformPartitionConvolution::FUniformPartitionConvolution	(	const FUniformPartitionConvolutionSettings &	InSettings,
		FSharedFFTRef	InFFTAlgorithm
	)

Create a Uniform Partition Convolution Algorithm

@params InSettings - Settings for algorithm.

Parameters

InFFTAlgorithm

- FFTAlgorithm used to transform inputs, outputs and impulse responses.

~FUniformPartitionConvolution()

Audio::FUniformPartitionConvolution::~FUniformPartitionConvolution ( )

virtual

Destructor.

Member Function Documentation

GetMatrixGain()

float Audio::FUniformPartitionConvolution::GetMatrixGain ( int32  InAudioInputIndex, int32  InImpulseResponseIndex, int32  InAudioOutputIndex  ) const

overridevirtual

Gets the gain between an audio input channel, impulse response and audio output channel.

([audio inputs] * [impulse responses]) x [gain matrix] = [audio outputs]

Gets the gain between an audio input, impulse response and audio output.

([audio inputs] * [impulse responses]) x [gain matrix] = [audio outputs]

Implements Audio::IConvolutionAlgorithm.

GetMaxNumImpulseResponseSamples()

int32 Audio::FUniformPartitionConvolution::GetMaxNumImpulseResponseSamples ( ) const

overridevirtual

Maximum supported length of impulse response.

Implements Audio::IConvolutionAlgorithm.

GetNumAudioInputs()

int32 Audio::FUniformPartitionConvolution::GetNumAudioInputs ( ) const

overridevirtual

Returns number of audio inputs.

Implements Audio::IConvolutionAlgorithm.

GetNumAudioOutputs()

int32 Audio::FUniformPartitionConvolution::GetNumAudioOutputs ( ) const

overridevirtual

Returns number of audio outputs.

Implements Audio::IConvolutionAlgorithm.

GetNumImpulseResponses()

int32 Audio::FUniformPartitionConvolution::GetNumImpulseResponses ( ) const

overridevirtual

Return the number of impulse responses.

Implements Audio::IConvolutionAlgorithm.

GetNumImpulseResponseSamples()

int32 Audio::FUniformPartitionConvolution::GetNumImpulseResponseSamples ( int32  InImpulseResponseIndex ) const

overridevirtual

Return the number of samples in an impulse response.

Implements Audio::IConvolutionAlgorithm.

GetNumSamplesInBlock()

int32 Audio::FUniformPartitionConvolution::GetNumSamplesInBlock ( ) const

overridevirtual

Returns the number of samples in an audio block.

Implements Audio::IConvolutionAlgorithm.

ProcessAudioBlock()

void Audio::FUniformPartitionConvolution::ProcessAudioBlock ( const float *const  InSamples[], float *const  OutSamples[]  )

overridevirtual

Process one block of audio.

InSamples is processed by the impulse responses. The output is placed in OutSamples.

@params InSamples - A 2D array of input deinterleaved audio samples. InSamples[GetNumAudioInputs()][GetNumSamplesInBlock()] @params OutSamples - A 2D array of output deinterleaved audio samples. OutSamples[GetNumAudioOutputs()][GetNumSamplesInBlock()]

Implements Audio::IConvolutionAlgorithm.

ResetAudioHistory()

void Audio::FUniformPartitionConvolution::ResetAudioHistory ( )

overridevirtual

Reset internal history buffer for all audio inputs.

Reset internal history buffers .

Implements Audio::IConvolutionAlgorithm.

SetImpulseResponse()

void Audio::FUniformPartitionConvolution::SetImpulseResponse ( int32  InImpulseResponseIndex, const float *  InSamples, int32  InNumSamples  )

overridevirtual

Set impulse response values.

Implements Audio::IConvolutionAlgorithm.

SetMatrixGain()

void Audio::FUniformPartitionConvolution::SetMatrixGain ( int32  InAudioInputIndex, int32  InImpulseResponseIndex, int32  InAudioOutputIndex, float  InGain  )

overridevirtual

Sets the gain between an audio input channel, impulse response and audio output channel.

([audio inputs] * [impulse responses]) x [gain matrix] = [audio outputs]

Sets the gain between an audio input, impulse response and audio output.

([audio inputs] * [impulse responses]) x [gain matrix] = [audio outputs]

Implements Audio::IConvolutionAlgorithm.

---

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

• Engine/Source/Runtime/SignalProcessing/Private/UniformPartitionConvolution.h
• Engine/Source/Runtime/SignalProcessing/Private/UniformPartitionConvolution.cpp