Class thl::dsp::dynamics::CompressorImpl#
-
class CompressorImpl : public thl::dsp::BaseProcessor#
Inheritance diagram for thl::dsp::dynamics::CompressorImpl:
Collaboration diagram for thl::dsp::dynamics::CompressorImpl:
A compressor and expander: downward and upward, with lookahead, sidechain and mix.
The key signal (the input, or the sidechain of process_with_sidechain()) passes an optional 12 dB/octave high-pass, its level is detected (LevelDetector: peak or RMS) and linked across the channels (StereoLink), the gain computer (GainComputer) turns it into a gain in dB and the smoother (GainSmoother: attack, hold, release, auto release) gives that gain its ballistics. The audio is delayed by the lookahead, so the gain moves before what it reacts to; latency_samples() is the lookahead (its maximum, k_max_lookahead_ms, is max_latency_samples()).
The output is the delayed input times (1 - mix) + mix · gain · makeup: a mix below 1 is parallel compression, with the dry part inside, delayed as the wet one. The compressor shifts no phase, so process_with_dry() is BaseProcessor’s (the input, undelayed, as its contract says). Auto makeup adds half the gain the curve gives at 0 dBFS, negated, to the makeup.
A derived class gives every Parameter through get_parameter_float(); all are read at the start of every process() call, at
modulation_offset(process_modulated() applies them at its change points). A NaN becomes the parameter’s default (k_parameter_ranges), the rest is clamped. Enums come as floats, rounded. Makeup and mix glide over k_ramp_seconds. Up to k_max_channels channels; further ones pass unchanged (delayed by the lookahead).The meters (any thread) hold the last block’s largest gain change, input and output peak.
Threads: prepare() allocates (audio stopped); process(), process_with_sidechain() and reset() are real-time safe.
Public Functions
-
virtual void prepare(const double &sample_rate, const size_t &samples_per_block, const size_t &num_channels) override#
-
virtual void process(thl::core::BufferView buffer, uint32_t modulation_offset = 0) override#
-
void process_with_sidechain(const thl::core::BufferView &buffer, const thl::core::BufferView &sidechain, uint32_t modulation_offset = 0)#
process() keyed by
sidechaininstead of the input: its channel c drives channel c (its last channel the ones beyond).An empty
sidechainkeys by the input. Same length asbuffer.
-
virtual void reset() override#
Clears the detectors, the gain, the sidechain filter and the lookahead delay; the glides end at their targets.
-
inline float gain_change_db() const#
The last block’s gain change farthest from 0 dB (negative: reduction), in dB.
-
inline float input_peak_db() const#
The last block’s input peak in dBFS (the key signal’s, before the high-pass).
-
inline float output_peak_db() const#
The last block’s output peak in dBFS.
Public Static Attributes
-
static constexpr size_t k_max_channels = 16#
-
static constexpr float k_max_lookahead_ms = 20.0f#
-
static constexpr double k_ramp_seconds = 0.02#
-
static constexpr float k_max_auto_makeup_db = 24.0f#
Protected Types
-
enum Parameter#
Values:
-
enumerator Threshold#
dB
-
enumerator Ratio#
up to infinity: limits (downward) or gates (expansion)
-
enumerator Knee#
dB
-
enumerator Range#
dB, the largest gain change
-
enumerator Mode#
DynamicsMode.
-
enumerator Attack#
ms
-
enumerator Hold#
ms
-
enumerator Release#
ms
-
enumerator AutoRelease#
0.5 on
-
enumerator Lookahead#
ms
-
enumerator Detector#
DetectorMode.
-
enumerator RmsWindow#
ms
-
enumerator Link#
StereoLink.
-
enumerator SidechainHighPass#
Hz; below 10 off, capped at 0.49 fs.
-
enumerator Makeup#
dB
-
enumerator AutoMakeup#
0.5 on
-
enumerator Mix#
0 dry to 1 wet
-
enumerator NumParameters#
-
enumerator Threshold#
Protected Functions
Protected Static Attributes
- static constexpr std::array< utils::ParameterRange, NumParameters > k_parameter_ranges {{{.m_default = -18.0f, .m_min = -80.0f, .m_max = 0.0f},{.m_default = 4.0f,.m_min = 1.0f,.m_max = std::numeric_limits<float>::infinity()},{.m_default = 6.0f, .m_min = 0.0f, .m_max = 24.0f},{.m_default = 60.0f, .m_min = 0.0f, .m_max = 60.0f},{.m_default = 0.0f, .m_min = 0.0f, .m_max = 3.0f},{.m_default = 10.0f, .m_min = 0.0f, .m_max = 500.0f},{.m_default = 0.0f, .m_min = 0.0f, .m_max = 500.0f},{.m_default = 100.0f, .m_min = 0.0f, .m_max = 5000.0f},{.m_default = 0.0f, .m_min = 0.0f, .m_max = 1.0f},{.m_default = 0.0f, .m_min = 0.0f, .m_max =k_max_lookahead_ms},{.m_default = 0.0f, .m_min = 0.0f, .m_max = 1.0f},{.m_default = 10.0f, .m_min = 0.1f, .m_max = 1000.0f},{.m_default = 0.0f, .m_min = 0.0f, .m_max = 2.0f},{.m_default = 0.0f, .m_min = 0.0f, .m_max = 20000.0f},{.m_default = 0.0f, .m_min = -24.0f, .m_max = 24.0f},{.m_default = 0.0f, .m_min = 0.0f, .m_max = 1.0f},{.m_default = 1.0f, .m_min = 0.0f, .m_max = 1.0f},}}
Every parameter’s default and range (enums: their values as floats), e.g.
for a host’s parameter definitions. A NaN reads as the default, the rest is clamped.
-
virtual void prepare(const double &sample_rate, const size_t &samples_per_block, const size_t &num_channels) override#