CV-to-scale quantizer on the 1 V/octave standard. More...

#include <scale_quantizer.hpp>

Public Member Functions
void	prepare (double sample_rate)
void	set_mode (QuantizeMode mode)
QuantizeMode	mode () const
void	set_edo (int n)
int	edo () const
void	set_scale_mask (std::uint16_t mask)
std::uint16_t	scale_mask () const
void	set_root_pc (int pc)
int	root_pc () const
void	set_hysteresis_cents (double cents)
double	hysteresis_cents () const
void	apply_reset_edge ()
	Verb 2: clears the hysteresis latch, so the next input quantizes on its own merits rather than being held by whatever the pattern left behind.
void	reset ()
int	latched_step () const
	The latched pre-mask step, in units of 1/N octave (edo mode) or semitones (scale_mask mode).
SampleType	process (SampleType cv_volts)
	Quantizes one sample of pitch CV, in volts.

Static Public Member Functions
static constexpr int	latency_samples ()

Static Public Attributes
static constexpr int	kDefaultEdo = 12
	Steps per octave in edo mode.
static constexpr int	kMaxEdo = 48
static constexpr std::uint16_t	kMajorMask = 0b0000'1010'1011'0101u
	Pitch classes of the major scale — bits 0, 2, 4, 5, 7, 9, 11.
static constexpr double	kHystCents = 20.0
	Extra travel, in cents, an input must make past a step boundary before the output follows it.
static constexpr double	kMaxHystCents = 1200.0
	Ceiling on the hysteresis setting, one octave.
static constexpr double	kMaxHystSteps = 0.45
	Ceiling on the hysteresis window expressed as a fraction of one step.
static constexpr double	kMaxAbsSteps = 4096.0
	Widest step index the quantizer will produce, in either direction.

Detailed Description

template<typename SampleType = float>
class pulp::signal::QuantizeScaleT< SampleType >

CV-to-scale quantizer on the 1 V/octave standard.

(informative) In 12-tone equal temperament the octave is twelve equal semitones of 100 cents, so on the 1 V/oct standard a pitch plays in tune when it is a multiple of 1/12 V — "steps of exactly 1/12 V = one semitone" [illustrated by public quantizer documentation, e.g. Doepfer A-156; 12-TET itself is common practice, cited for illustration only]. The pitch-class bitmask convention is shared verbatim with the MIDI series' scale lock, so a scale authored on either side means the same thing here.

Hysteresis is the load-bearing part. A quantizer without it chatters whenever its input hovers on a step boundary — an LFO peak, a slow envelope, a noisy CV — and the chatter is audible as a trill, not as jitter. A step change therefore requires the input to cross the boundary by hyst_cents, and the latch is the block's only state.

The latch holds the chromatic (pre-mask) step rather than the snapped output. That is what makes the hysteresis window a fixed number of cents of input travel: if it latched the output, the window would silently widen and narrow with the gaps in the scale, so a semitone-wide leading tone and a minor-third gap would debounce differently.

Glide is not baked in. The output is a stepped voltage; a caller wanting smooth transitions puts a SlewLimiterT (kit) downstream, which keeps this block pure and lets the same quantizer feed a glide and a hard-stepped path.

USE. Taming the rungler or the Cartesian walk — any wild CV becomes an in-key melody. Sample-and-hold pitch — a SampleHoldT (kit) on an LFO into the quantizer into an oscillator is the classic random arpeggio. Microtonal — EDO-N mode reaches 24-tone quarter tones and 19- or 31-tone tunings without a lookup table.

RT contract: as the header. One llround-free floor and a bounded search.