DSP API Reference
All DSP modules live in the ovtdsp:: namespace (chosen instead of dsp
to avoid ambiguity with juce::dsp). This page documents the real public
classes/methods as found in the headers under Source/dsp/.
Pitch detection
ovtdsp::IPitchDetector (interface)
| Member |
Description |
virtual void prepare(double sampleRate, int blockSize) |
Initialize with sample rate and block size. |
virtual void reset() |
Clear internal state. |
virtual float detectPitch(const float* samples, int numSamples) |
Return fundamental frequency in Hz, or 0.0 if unvoiced. |
virtual void setThreshold(float t) |
Detection sensitivity (0.01–0.99; lower = stricter). |
virtual float getThreshold() const |
Current threshold. |
virtual juce::String getName() const |
Human-readable algorithm name. |
ovtdsp::YinPitchDetector : public IPitchDetector (active)
YIN implementation (de Cheveigne & Kawahara). Extras over the interface:
void setFrequencyRange(float minHz, float maxHz) — narrows the search range
(used by the Voice Type selector). Safe to call from the audio thread.
Legacy
ovtdsp::PitchDetector (in PitchDetector.h) is a legacy YIN detector
superseded by YinPitchDetector and is no longer used by the pipeline.
Quantization
ovtdsp::ScaleQuantizer
| Member |
Description |
void setKey(int keyInSemitones) |
Set tonic (0=C … 11=B). |
void setScale(ovtdsp::Scale scale) |
Set mode/scale (Chromatic=0 … Custom=13). |
void setCustomIntervals(const juce::Array<int>&) |
Set the 12 custom notes (no key offset); ignored unless Scale::Custom. |
float quantize(float hzIn) const |
Return the nearest in-scale Hz, or hzIn if none. |
const juce::Array<int>& getScaleIntervals() const |
Active scale notes (semitones 0..11 relative to C). |
bool isCustom() const |
Whether Scale::Custom is active. |
Free helpers:
float hzToSemitones(float hz) / float semitonesToHz(float semi) — A4 = 440 Hz.
ovtdsp::Scale (enum)
Chromatic=0, Major, MelodicMinor, HarmonicMinor, NaturalMinor, MajorPentatonic,
MinorPentatonic, Blues, Dorian, Phrygian, Lydian, Mixolydian, Locrian, Custom=13.
Pitch shifting
ovtdsp::IPitchShifter (interface)
| Member |
Description |
virtual void prepare(double sampleRate, int maximumBlockSize) |
Initialize. |
virtual void reset() |
Flush internal state (FIFOs). |
virtual void process(juce::AudioBuffer<float>&, float ratio, float f0) |
Process in place (ratio = transposition, f0 needed by PSOLA). |
virtual int getLatencySamples() const |
Internal latency in samples. |
ovtdsp::PitchShifter : public IPitchShifter (PSOLA)
PSOLA is the only pitch shifter. Public members:
| Member |
Description |
void process(buffer, float pitchRatio, float formantRatio, float f0) |
In-place process with a distinct formant ratio. |
void process(input, output, float pitchRatio, float formantRatio, float f0) |
Out-of-place variant. |
void setLatencyMs(float) |
Set latency, clamped to 8–40 ms. |
int getLatencySamples() const |
Current latency in samples. |
void setAttackTimeMs(float) |
Voice-onset attack time (default 30 ms). |
void setAttackEnvelopeEnabled(bool) |
Enable/disable internal onset fade (coordinate with an external helper). |
void setExternalAttackGain(float gain, float blockDurSec) |
External block-level attack-gain driver (negative = disable). |
void setExternalAttackTauSeconds(float) |
TC of the external attack smoother (default 0.015 s). |
void resetExternalAttackGain() |
Snap smoother to 1.0 and disable external driver. |
void resetSoft() |
Reset state without re-arming the startup fade-in (session seek/preset). |
void forceCreateTestGrain() |
Debug hook to force a single test grain. |
Speed / retarget smoothing
ovtdsp::RetargetEnvelope
| Member |
Description |
void prepare(double sampleRate) |
Initialize. |
void reset() |
Reset state. |
void setSpeed(float ms) |
Retarget time in milliseconds. |
float processSample(float targetRatio) |
Smooth a single sample of ratio. |
float processBlock(float targetRatio, int numSamples) |
Buffer-size-independent block smoothing (recommended). |
| Member |
Description |
enum class Mode |
Legacy, MultiFormant (F1–F4), Allpass (F1–F4 allpass cascade). |
enum class Strategy |
Current (partial 1/sqrt(r), male-default centers), P0 (full 1/r, voice-type-aware centers). |
void prepare(double, int) / void reset() |
Lifecycle. |
void process(juce::AudioBuffer<float>&, float ratio) |
Apply formant shift (must run before PSOLA). |
setEnabled/isEnabled, setFormantShift(float semitones) |
Enable + manual shift in semitones. |
setMode/getMode, setStrategy/getStrategy |
Mode/strategy selectors. |
setQMultiplier(float), setSmoothingAlpha(float) |
Resonance Q and biquad-coefficient smoothing. |
setVoiceType(int) |
Voice type (0=Universal…5=Soprano) for P0 centers. |
setFormant(int index, float freqHz, float q, float gainDb) |
Configure a specific formant (MultiFormant only). |
LPC cross-synthesis formant preservation (the "exact" approach).
| Member |
Description |
enum class Mode |
C0 (P1: plain LPC cross-synthesis), C1Hybrid (P2: pre-emphasis + LPC interpolation + hybrid fallback). |
explicit LpcFormantPreserver(int order = 18) |
Construct with LPC order. |
void prepare(double, int) / void reset() |
Lifecycle. |
void setOrder(int) |
Change the LPC order. |
void process(juce::AudioBuffer<float>& out, const juce::AudioBuffer<float>& reference, float ratio, Mode) |
Re-color out (post-shift) using reference (pre-shift) formants, in place. |
Harmony
ovtdsp::HarmonyEngine
| Member |
Description |
enum class HarmonyType |
None + 21 types (ThirdBelow, ThirdAbove, ThirdBelowAbove, Fourth*, Fifth*, Octave*, VocalStack3/4, PowerChord, ParallelThird, Drone, Unison2, UnisonOctaves4). |
void prepare(double sampleRate) |
Store sample rate for oscillator increments. |
juce::Array<float> getHarmonyNotes(float baseFreq, const juce::Array<int>& scaleIntervals, HarmonyType) const |
Compute per-voice frequencies. |
int getScaleDegree(float freq, int key, const juce::Array<int>&) const |
Degree of a frequency within the scale. |
void renderHarmonies(inputFreq, harmonyFrequencies, volume, sampleRate, outputBuffer, key, scaleIndex, blend, toneMode, toneColor, gateActive) |
Synthesize voices into the buffer. |
static juce::String getHarmonyName(HarmonyType) |
Display name. |
static int getHarmonyVoiceCount(HarmonyType) |
Number of generated voices. |
void setVoiceGate(bool) |
Voice gating (note on/off). |
void setEnvelopeTimes(float attackMs, float releaseMs) |
Per-voice attack/release. |
bool isActive() const |
Whether any voice is sounding/releasing. |
Effects
ovtdsp::IEffect (interface)
| Member |
Description |
virtual juce::String getId() const |
Unique id (e.g. "reverb"). |
virtual juce::String getName() const |
Human-readable name (e.g. "Reverb"). |
virtual void prepare(double sampleRate, int maximumBlockSize) |
Prepare. |
virtual void reset() |
Flush delays/reverb tails. |
virtual void process(juce::AudioBuffer<float>& buffer, bool enabled, float wetMix) |
Apply in place. |
ovtdsp::ReverbEffect : public IEffect
setRoomSize(float) / setDamping(float) — reverb tuning.- Wraps
juce::Reverb; smooths the master enable gain to avoid clicks.
ovtdsp::NoiseGate
void prepare(double sr), setEnabled(bool), setThresholdDb(float),
bool isEnabled(), float getCurrentGain(), void process(juce::AudioBuffer<float>&).- RMS-based, sample-accurate, with hysteresis and smooth attack/release.
Graphic-mode curve
ovtdsp::PitchPoint (struct)
{ double time; float pitch; } — time in seconds, pitch in Hz, with
==/!= operators.
ovtdsp::PitchCurve
Sorted list of PitchPoint with linear interpolation between points
(evaluated by binary search, O(log N)).
| Member |
Description |
int addOrUpdatePoint(double time, float pitch) |
Add or replace nearest point; returns its index. |
bool removePointNear(double time, double toleranceSec = 0.05) |
Remove nearest point within tolerance. |
void clear() / int getNumPoints() |
Reset / count. |
const PitchPoint& getPoint(int) const / PitchPoint& getPoint(int) |
Direct access. |
void setPointPitch(int, float) / setPointTimeAndPitch(int&, double, float) / setMultiplePointsTimeAndPitch(...) |
Editing helpers (keep time-sorted). |
float getPitchAt(double time, float defaultValue = 0.0f) const |
Evaluate the curve at time. |
static float snapToIntervals(float hz, const juce::Array<int>& intervals) |
Snap to an explicit interval set. |
std::unique_ptr<juce::XmlElement> toXml() const / void fromXml(const juce::XmlElement&) |
Serialization. |
void loadPreset(const juce::String& presetName) |
Load a factory curve preset ("default", "spoken", "lyric", "rap", "robot"). |
setStepMode/setSnapEnabled/setSnapToGridEnabled + getters |
Editing state (persisted with the curve). |
Preset morphing
ovtdsp::MorphState (struct)
Snapshot of all interpolable plugin parameters (continuous floats, discrete
ints, booleans) plus a PitchCurve and a name.
ovtdsp morph helpers (inline)
| Function |
Description |
MorphState captureState(juce::AudioProcessorValueTreeState&, const ovtdsp::PitchCurve&, const juce::String& name) |
Capture all parameters into a state. |
juce::StringArray getMorphParameterIds() |
All parameter IDs a morph can drive. |
void applyInterpolatedState(params, source, target, float morphAmount, const juce::StringArray* exclude) |
Apply interpolation; continuous lerp, discrete/bool step at 50%; exclude skips externally-automated params. |
Note utilities
ovtdsp::NoteUtils.h helpers
| Function |
Description |
float hzToMidiFloat(float hz) |
Hz → MIDI float. |
float midiToHz(float midi) |
MIDI → Hz. |
float hzToCents(float hzDetected, float hzTarget) |
Cents offset (positive = detected above target). |
int midiToOctave(int) / int midiToNoteInOctave(int) |
Octave / note-in-octave. |
const char* noteInOctaveName(int) |
Short name (C, C#, …, B). |
juce::String hzToNoteName(float hz) |
Full name (e.g. "F3"), "--" if invalid. |
struct NoteInfo |
name, targetName, midi, targetMidi, octave, noteInOct, cents, valid. |
NoteInfo describePitch(float hzIn, float hzTarget) |
Note + cents display data used by the visualizer. |