dspark::Oversampling< T, MaxChannels > Class Template Reference

Power-of-two oversampling processor with polyphase anti-aliasing. More...

#include <Oversampling.h>

Public Types
enum class	Quality { Low , Medium , High , Maximum }
	Quality presets defining the steepness and rejection of the anti-aliasing filter. More...

Public Member Functions
	Oversampling (int factor=2, Quality quality=Quality::High)
	Constructs the oversampling engine.
void	prepare (const AudioSpec &spec)
	Pre-allocates buffers and computes FIR filter coefficients.
void	reset () noexcept
	Flushes all internal delay lines and history buffers. Call this when resetting the transport or to clear audio tails.
int	getFactor () const noexcept
	Returns the current oversampling factor.
Quality	getQuality () const noexcept
	Returns the current quality preset.
int	getLatency () const noexcept
	Calculates the exact group delay latency of the oversampling chain.
AudioBufferView< T >	upsample (AudioBufferView< const T > input) noexcept
	Upsamples the input base-rate signal into the internal high-rate buffer.
void	downsample (AudioBufferView< T > output) noexcept
	Downsamples the internal high-rate buffer back to base-rate.

Detailed Description

template<typename T, int MaxChannels = 16>
class dspark::Oversampling< T, MaxChannels >

Power-of-two oversampling processor with polyphase anti-aliasing.

Usage is a paired streaming operation per audio callback:

auto up = os.upsample(block);   // 1. base rate -> high rate
process(up);                    // 2. nonlinear processing at the high rate
os.downsample(block);           // 3. high rate -> base rate, in place

Note

upsample() and downsample() form a pair: downsample() consumes the high-rate samples the most recent upsample() left in the internal buffer, so call them once each per block, with the same block length.

Threading: prepare() and reset() belong to the setup thread (prepare allocates). upsample() and downsample() belong to the audio thread and are noexcept and allocation-free. There are no runtime setters – factor and quality are fixed at construction and all filter state is only touched by the audio thread – so no cross-thread synchronisation is needed.

Template Parameters

T	Sample type (float or double).
MaxChannels	Maximum number of channels supported.

Definition at line 67 of file Oversampling.h.

Member Enumeration Documentation

Quality

template<typename T , int MaxChannels = 16>

enum class dspark::Oversampling::Quality

strong

Quality presets defining the steepness and rejection of the anti-aliasing filter.

Enumerator
Low	31 taps/stage, ~-40 dB stopband. For fast previewing.
Medium	63 taps/stage, ~-60 dB stopband. General purpose.
High	127 taps/stage, ~-80 dB stopband. Professional standard.
Maximum	255 taps/stage, ~-100 dB stopband. Mastering grade.

Definition at line 73 of file Oversampling.h.

Constructor & Destructor Documentation

Oversampling()

template<typename T , int MaxChannels = 16>

dspark::Oversampling< T, MaxChannels >::Oversampling ( int  factor = 2, Quality  quality = Quality::High  )

inlineexplicit

Constructs the oversampling engine.

Parameters

factor	Oversampling factor. Must be a power of two (1, 2, 4, 8, 16).
quality	Filter quality preset. Default is High (-80 dB).

Precondition

factor >= 1 and is a power of 2.

Note

Release builds normalise an invalid factor into range and round it down to a power of two (debug builds assert).

Definition at line 89 of file Oversampling.h.

Member Function Documentation

downsample()

template<typename T , int MaxChannels = 16>

void dspark::Oversampling< T, MaxChannels >::downsample ( AudioBufferView< T >  output )

inlinenoexcept

Downsamples the internal high-rate buffer back to base-rate.

Consumes the high-rate samples written by the most recent upsample() (and modified in place by the caller in between).

Parameters

output

Buffer view where the base-rate samples will be written. Use the same block length as the paired upsample() input.

Note

Real-time safe.

Definition at line 212 of file Oversampling.h.

getFactor()

template<typename T , int MaxChannels = 16>

int dspark::Oversampling< T, MaxChannels >::getFactor ( ) const

inlinenoexcept

Returns the current oversampling factor.

Definition at line 142 of file Oversampling.h.

getLatency()

template<typename T , int MaxChannels = 16>

int dspark::Oversampling< T, MaxChannels >::getLatency ( ) const

inlinenoexcept

Calculates the exact group delay latency of the oversampling chain.

Returns

Latency in samples at the BASE (1x) sample rate.

Definition at line 151 of file Oversampling.h.

getQuality()

template<typename T , int MaxChannels = 16>

Quality dspark::Oversampling< T, MaxChannels >::getQuality ( ) const

inlinenoexcept

Returns the current quality preset.

Definition at line 145 of file Oversampling.h.

prepare()

template<typename T , int MaxChannels = 16>

void dspark::Oversampling< T, MaxChannels >::prepare ( const AudioSpec &  spec )

inline

Pre-allocates buffers and computes FIR filter coefficients.

Parameters

spec	The audio specification at the base (1x) sample rate.

Postcondition

The processor is ready to handle up to spec.maxBlockSize base samples.

Definition at line 112 of file Oversampling.h.

reset()

template<typename T , int MaxChannels = 16>

void dspark::Oversampling< T, MaxChannels >::reset ( )

inlinenoexcept

Flushes all internal delay lines and history buffers. Call this when resetting the transport or to clear audio tails.

Definition at line 134 of file Oversampling.h.

upsample()

template<typename T , int MaxChannels = 16>

AudioBufferView< T > dspark::Oversampling< T, MaxChannels >::upsample ( AudioBufferView< const T >  input )

inlinenoexcept

Upsamples the input base-rate signal into the internal high-rate buffer.

Parameters

input

View of the base-rate audio buffer. Blocks longer than prepare()'s maxBlockSize are truncated (release-safe); channels beyond the prepared channel count are ignored.

Returns

A view of the internal oversampled buffer (length = input samples * factor).

Note

Real-time safe. Pair every call with one downsample() of the same block length before the next upsample().

Definition at line 174 of file Oversampling.h.

---

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

• Core/Oversampling.h