A DSP processor defined by a user-provided function. More...

#include <hart_dsp_function.hpp>

Inheritance diagram for DSPFunction< SampleType >:

Public Member Functions
	DSPFunction (std::function< SampleType(SampleType)> dspFunction, const std::string &label={})
	Creates a DSP instance based on a sample-wise DSP function.

	DSPFunction (std::function< void(AudioBuffer< SampleType > &)> dspFunction, const std::string &label={})
	Creates a DSP instance based on a block-wise replacing (in-place) DSP function.

	DSPFunction (std::function< void(const AudioBuffer< SampleType > &, AudioBuffer< SampleType > &)> dspFunction, const std::string &label={})
	Creates a DSP instance based on a block-wise non-replacing (separate input and output buffers) DSP function.

void	prepare (double sampleRateHz, size_t numInputChannels, size_t numOutputChannels, size_t maxBlockSizeFrames) override
	Prepare for processing.

void	process (const AudioBuffer< SampleType > &input, AudioBuffer< SampleType > &output, const EnvelopeBuffers &, ChannelFlags) override
	Processes the audio.

void	represent (std::ostream &stream) const override
	Makes a text representation of this DSP effect for test failure outputs.

void	reset () override
	Resets to initial state.

void	setValue (int, double) override
	Sets DSP value.

bool	supportsSampleRate (double) const override
	Tells whether this effect supports given sample rate.

bool	supportsEnvelopeFor (int) const override
	Tells whether this effect accepts automation envelopes for a particular parameter.

bool	supportsChannelLayout (size_t numInputChannels, size_t numOutputChannels) const override
	Tells the runner (host) whether this effect supports a specific i/o configuration.

Public Member Functions inherited from DSP< SampleType, DSPFunction< SampleType > >
DSPFunction< SampleType > &	withEnvelope (int paramId, Envelope &&envelope) &
	Adds envelope to a specific parameter by moving it.

DSPFunction< SampleType > &&	withEnvelope (int paramId, Envelope &&envelope) &&
	Adds envelope to a specific parameter by moving it.

DSPFunction< SampleType > &	withEnvelope (int paramId, const Envelope &envelope) &
	Adds envelope to a specific parameter by copying it.

DSPFunction< SampleType > &&	withEnvelope (int paramId, const Envelope &envelope) &&
	Adds envelope to a specific parameter by copying it.

virtual std::unique_ptr< DSPBase< SampleType > >	move () override
	Returns a smart pointer with a moved instance of this object.

DSPFunction< SampleType > &	atChannels (std::initializer_list< size_t > channelsToProcess) &
	Makes this DSP process only specific channels, and ignore the rest.

DSPFunction< SampleType > &&	atChannels (std::initializer_list< size_t > channelsToProcess) &&
	Makes this DSP process only specific channels, and ignore the rest.

DSPFunction< SampleType > &	atChannel (size_t channelToProcess) &
	Makes this DSP process only specific channels, and bypass the rest.

DSPFunction< SampleType > &&	atChannel (size_t channelToProcess) &&
	Makes this DSP process only specific channels, and bypass the rest.

DSPFunction< SampleType > &	atAllChannels () &
	Makes this DSP apply to all channels.

DSPFunction< SampleType > &&	atAllChannels () &&
	Makes this DSP apply to all channels.

DSPFunction< SampleType > &	atAllChannelsExcept (std::initializer_list< size_t > channelsToSkip) &
	Makes this DSP process only specific channels, and bypass the rest.

DSPFunction< SampleType > &&	atAllChannelsExcept (std::initializer_list< size_t > channelsToSkip) &&
	Makes this DSP process only specific channels, and bypass the rest.

Public Member Functions inherited from DSPBase< SampleType >
virtual double	getValue (int) const
	Retrieves DSP value.

virtual std::unique_ptr< DSPBase< SampleType > >	copy () const
	Returns a smart pointer with a copy of this object.

virtual	~DSPBase ()=default
	Destructor.

	DSPBase ()=default
	Default constructor.

	DSPBase (const DSPBase &other)
	Copies from another DSP effect instance.

	DSPBase (DSPBase &&other) noexcept
	Move constructor.

DSPBase &	operator= (const DSPBase &other)
	Copies from another DSP effect instance.

DSPBase &	operator= (DSPBase &&other) noexcept
	Move assignment.

bool	hasEnvelopeFor (int paramId)
	Checks if there's an automation envelope attached to a specific parameter.

void	prepareWithEnvelopes (double sampleRateHz, size_t numInputChannels, size_t numOutputChannels, size_t maxBlockSizeFrames)
	Prepares all the attached envelopes and the effect itself for processing.

void	resetWithEnvelopes ()
	Resets the DSP instance, and all associated Envelopes.

ChannelFlags	getChannelsToProcess ()
	Returns a structure indicating which channels should be processed by this DSP.

void	processWithEnvelopes (const AudioBuffer< SampleType > &input, AudioBuffer< SampleType > &output)
	Renders all the automation envelopes and processes the audio.

void	representWithActiveChannels (std::ostream &stream) const
	Makes a text representation of this DSP with optional "atChannels" appendix.

Additional Inherited Members
Protected Attributes inherited from DSPBase< SampleType >
std::unordered_map< int, std::unique_ptr< Envelope > >	m_envelopes

ChannelFlags	m_channelsToProcess {true}

Related Symbols inherited from DSP< SampleType, DSPFunction< SampleType > >
std::ostream &	operator<< (std::ostream &stream, const DSPBase< SampleType > &dsp)
	Prints readable text representation of the DSP object into the I/O stream.

Detailed Description

template<typename SampleType>
class hart::DSPFunction< SampleType >

A DSP processor defined by a user-provided function.

It allows defining audio processing logic using a lambda or function object, without defining a dedicated DSP subclass.

Three function styles are supported:

1. Sample-wise processing: SampleType (SampleType value)

hart::DSPFunction
A DSP processor defined by a user-provided function.
Definition hart_dsp_function.hpp:62

Processes each sample independently.

2. Block-wise replacing (in-place) processing: void (AudioBuffer<SampleType>& buffer)

Processes the buffer in-place. The buffer is pre-filled with input data.

3. Block-wise non-replacing processing: void (const AudioBuffer<SampleType>& input, AudioBuffer<SampleType>& output)

Provides separate input and output buffers.

Buffer invariants

For mutable buffers in block-wise replacing and non-replacing modes:

Number of frames must not be changed
Number of channels must not be changed
Sample rate must not be changed

Disobeying those constraints will result in a runtime error.

Example: auto myDsp = DSPFunction<float> (

[] (float x) { return x * 0.5f; },

"Gain"

);

Note: This class is primarily intended for simple and expressive DSP definitions in tests. For more complex or stateful processors, it's encouraged to implement a dedicated DSP subclass.

Definition at line 60 of file hart_dsp_function.hpp.

Constructor & Destructor Documentation

◆ DSPFunction() [1/3]

template<typename SampleType >

DSPFunction	(	std::function< SampleType(SampleType)>	dspFunction,
		const std::string &	label = `{}`
	)

inline

Creates a DSP instance based on a sample-wise DSP function.

Note: If you need to be aware of sample rate value, or what channel a given sample is from, use block-wise function signature instead (see other ctors of this class for details). Or define a full-featured DSP subclass for your DSP, instead of a simple function-based one.

This is the simplest form of a custom DSP. The function is applied independently to each sample in the input signal.

Example: auto myDsp = DSPFunction<float> (

[] (float x) { return x * 0.5f; },

"Gain"

);

Parameters

dspFunction	Function with signature: SampleType (SampleType value) Will be called to process each sample, expected to return a processed value.
label	Optional human-readable label to use in test error reporting.

Definition at line 86 of file hart_dsp_function.hpp.

◆ DSPFunction() [2/3]

template<typename SampleType >

DSPFunction	(	std::function< void(AudioBuffer< SampleType > &)>	dspFunction,
		const std::string &	label = `{}`
	)

inline

Creates a DSP instance based on a block-wise replacing (in-place) DSP function.

The provided function will be called for each block of input audio. It will receive a buffer pre-filled with input audio data. The function is expected to modify the buffer directly. The provided buffer is guaranteed to have:

A number of channels equal to max (inputChannels, outputChannels)
Input channels populated with input data
Additional channels (if any) initialized to zero
Number of frames equal to the block size specified in the test case, or less than that in case of partial blocks.
Correct sample rate value in its metadata (see AudioBuffer::getSampleRateHz())

For the provided buffer, the function must not change:

Number of channels
Number of frames
Sample rate

Disobeying those constraints will result in a runtime error.

Example: auto myDsp = DSPFunction<float> (

[] (auto& buffer)

{

// Process buffer in-place

},

"My DSP function"

);

Parameters

dspFunction	Function with signature: void (AudioBuffer<SampleType>& buffer) Expected to read the input samples in the buffer and replace them with processed data.
label	Optional human-readable label to use in test error reporting.

Definition at line 124 of file hart_dsp_function.hpp.

◆ DSPFunction() [3/3]

template<typename SampleType >

DSPFunction	(	std::function< void(const AudioBuffer< SampleType > &, AudioBuffer< SampleType > &)>	dspFunction,
		const std::string &	label = `{}`
	)

inline

Creates a DSP instance based on a block-wise non-replacing (separate input and output buffers) DSP function.

This form provides full control over how output is generated from input. The output buffer is pre-allocated with the expected size.

For the output buffer, the function must not change:

Number of channels
Number of frames
Sample rate

Disobeying those constraints will result in a runtime error.

Example: auto myDsp = DSPFunction<float> (

[] (const auto& in, auto& out)

{

// Fill output using input

},

"My DSP function"

);

Parameters

dspFunction	Function with signature: void (const AudioBuffer<SampleType>& input, AudioBuffer<SampleType>& output) Expected to fill the output buffer with processed data from the input buffer.
label	Optional human-readable label to use in test error reporting.

Definition at line 158 of file hart_dsp_function.hpp.

Member Function Documentation

◆ prepare()

template<typename SampleType >

void prepare	(	double	sampleRateHz,
		size_t	numInputChannels,
		size_t	numOutputChannels,
		size_t	maxBlockSizeFrames
	)

inlineoverridevirtual

Prepare for processing.

In real-time DSP, such methods are usually used for allocating memory and other non-realtime-safe and heavyweight operations. But keep in mind that that HART does not do real-time processing, so this merely follows common real-time DSP design conventions, where non-realtime operations are done in a separate callback like this one. This method is guaranteed to be called after supportsChannelLayout() and supportsSampleRate(), but before process(). It is guaranteed that the number of input and output channels obeys supportsChannelLayout() and supportsSampleRate() preferences. It is guaranteed that all subsequent process() calls will be in line with the arguments received in this callback.

Parameters

sampleRateHz	sample rate at which the audio should be interpreted and processed
numInputChannels	Number of input channels
numOutputChannels	Number of output channels
maxBlockSizeFrames	Maximum block size in frames (samples)

Implements DSPBase< SampleType >.

Definition at line 163 of file hart_dsp_function.hpp.

◆ process()

template<typename SampleType >

void process	(	const AudioBuffer< SampleType > &	input,
		AudioBuffer< SampleType > &	output,
		const EnvelopeBuffers &	envelopeBuffers,
		ChannelFlags	channelsToProcess
	)

inlineoverridevirtual

Processes the audio.

Depending on circumstances, this callback will either be called once to process an entire piece of audio from start to finish, or called repeatedly, one block at a time (see AudioTestBuilder::withBlockSize()). All audio blocks except the last one are guaranteed to be equal to maxBlockSizeFrames set in prepare() callback. It is guaranteed that input and output buffers are equal in length in frames (samples) to each, but they might have different number of channels. Use supportsChannelLayout() to indicate whether the effect supports a specific i/o configuration or not, as it will be called before prepare(). It is guaranteed that envelopeBuffers will only contain the values for all attached envelopes for this instance of DSP effect, and will not contain any data (including key with empty item) if there's no envelope attached to a specific parameter ID in this effects's instance. It will never contain envelopes for IDs that get rejected by supportsEnvelopeFor(). Each vector in the envelopeBuffers map is guaranteed to be at least as large as input (and output) block. For partial blocks, only the first input.getNumFrames() elements contain valid data for the current block; the rest may be stale or uninitialized. DSP implementations are expected to ignore values beyond that index range.

Note: This method may be called in a replacing manner, i. e. input and output may be references to the same object.

Warning: Remember that the very last block of audio is almost always smaller than the block size set in prepare(), so be careful with buffer bounds.

Note: Note that this method does not have to be real-time safe, as all rendering always happens offline. Also note that, unlike real-time audio applications, this method is called on the same thread as all others like prepare().

Parameters

input	Input audio block
output	Output audio block
envelopeBuffers	Envelope values for this block, see EnvelopeBuffers
channelsToProcess	Channels that need processing. Process channels that are marked `true`, bypass ones marked `false`.

Implements DSPBase< SampleType >.

Definition at line 172 of file hart_dsp_function.hpp.

◆ represent()

template<typename SampleType >

void represent ( std::ostream & stream ) const

inlineoverridevirtual

Makes a text representation of this DSP effect for test failure outputs.

It is strongly encouraged to follow python's repr() conventions for returned text - basically, put something like "MyClass(value1, value2)" (with no quotes) into the stream whenever possible, or "<Readable info in angled brackets>" otherwise. Also, use built-in stream manipulators like dbPrecision wherever applicable. Use HART_DEFINE_GENERIC_REPRESENT() to get a basic implementation for this method.

Parameters

[out] stream Output stream to write to

Implements DSPBase< SampleType >.

Definition at line 195 of file hart_dsp_function.hpp.

◆ reset()

template<typename SampleType >

void reset ( )

inlineoverridevirtual

Resets to initial state.

Ideally should be implemented in a way that audio produced after resetting is identical to audio produced after instantiation

Reimplemented from DSPBase< SampleType >.

Definition at line 200 of file hart_dsp_function.hpp.

◆ setValue()

template<typename SampleType >

void setValue	(	int	paramId,
		double	value
	)

inlineoverridevirtual

Sets DSP value.

Parameters

paramId	Some ID that your subclass understands; use of enums is encouraged for readability
value	Value of the param in an appropriate unit; use of SI units is encouraged (i.e. s instead of ms. Hz instead of kHz) to make better use of unit literals (see Units)

Warning: This method is only called to set a fixed value before processing, and is not called to do automation (via hart::Envelope) If you want your class to support automation for a specific parameter, override supportsEnvelopeFor(), and then use envelopeBuffers provided in process() callback.

Implements DSPBase< SampleType >.

Definition at line 201 of file hart_dsp_function.hpp.

◆ supportsSampleRate()

template<typename SampleType >

bool supportsSampleRate ( double ) const

inlineoverridevirtual

Tells whether this effect supports given sample rate.

It is guaranteed to be called before prepare()

Parameters

sampleRateHz Sample rate in question

Returns: true if effect is capable of interpreting and processing in a given sample rate, false otherwise

Reimplemented from DSPBase< SampleType >.

Definition at line 202 of file hart_dsp_function.hpp.

◆ supportsEnvelopeFor()

template<typename SampleType >

bool supportsEnvelopeFor ( int ) const

inlineoverridevirtual

Tells whether this effect accepts automation envelopes for a particular parameter.

Parameters

paramId Some ID that your subclass understands

Returns: true if your subclass can process automation for this parameter, false otherwise

Reimplemented from DSPBase< SampleType >.

Definition at line 203 of file hart_dsp_function.hpp.

◆ supportsChannelLayout()

template<typename SampleType >

bool supportsChannelLayout	(	size_t	numInputChannels,
		size_t	numOutputChannels
	)		const

inlineoverridevirtual

Tells the runner (host) whether this effect supports a specific i/o configuration.

It is guaranteed that the effect will not receive unsupported number of channels in process(). However, it is not always possible to handle gracefully channel layout being unsupported, so in some circumstances it can cause an exception or a test failure. This method is guaranteed to be called at least once before prepare()

Implements DSPBase< SampleType >.

Definition at line 205 of file hart_dsp_function.hpp.

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

include/dsp/hart_dsp_function.hpp

Public Member Functions

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ DSPFunction() [1/3]

◆ DSPFunction() [2/3]

◆ DSPFunction() [3/3]

Member Function Documentation

◆ prepare()

◆ process()

◆ represent()

◆ reset()

◆ setValue()

◆ supportsSampleRate()

◆ supportsEnvelopeFor()

◆ supportsChannelLayout()