homeburger/metabalus
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

polyphony checkpoint

Browse Source
This commit is contained in:
Bliblank
2026-01-11 23:00:44 -06:00
parent c0c914b40d
commit 1c9acb2fd9
6 changed files with 223 additions and 119 deletions
Download Patch File Download Diff File Expand all files Collapse all files

115
src/synth/Synth.cpp
View File

@@ -25,48 +25,41 @@ void Synth::updateParams() {
void Synth::setSampleRate(uint32_t sampleRate) {
sampleRate_ = sampleRate;
filter_.setSampleRate(static_cast<float>(sampleRate));
for(Voice& v : voices_) {
v.setSampleRate(static_cast<float>(sampleRate));
}
}
inline float Synth::getParam(ParamId id) {
return params_[static_cast<size_t>(id)].current;
}
inline float Synth::noteToFrequency(uint8_t note) {
return SYNTH_PITCH_STANDARD * pow(2.0f, static_cast<float>(note - SYNTH_MIDI_HOME) / static_cast<float>(SYNTH_NOTES_PER_OCTAVE));
}
void Synth::handleNoteEvent(const NoteEvent& event) {
if(event.type == NoteEventType::NoteOn) {
// add note to activeNotes list
if (std::find(heldNotes_.begin(), heldNotes_.end(), event.note) == heldNotes_.end()) {
heldNotes_.push_back(event.note);
gainEnvelope_.noteOn();
cutoffEnvelope_.noteOn();
resonanceEnvelope_.noteOn();
// TODO: envelopes in an array so we can loop over them
// TODO: find quietest voice and assign a note to it instead of just the first inactive one
// find inactive voice and start it with the given note
for(Voice& v : voices_) {
if(!v.isActive()) {
v.noteOn(event.note, event.velocity);
break;
}
}
} else {
// remove note from activeNotes list
auto it = std::find(heldNotes_.begin(), heldNotes_.end(), event.note);
if (it != heldNotes_.end()) {
heldNotes_.erase(it);
// find voice associated with note event and end it
for(Voice& v : voices_) {
if(v.isActive() && v.note() == event.note) {
v.noteOff();
break;
}
}
}
updateCurrentNote();
}
void Synth::updateCurrentNote() {
if(heldNotes_.empty()) {
gainEnvelope_.noteOff(); // TODO: move somewhere else when polyphony
cutoffEnvelope_.noteOff();
resonanceEnvelope_.noteOff();
return;
}
uint8_t note = heldNotes_.back();
frequency_ = noteToFrequency(note);
}
void Synth::process(float* out, uint32_t nFrames, uint32_t sampleRate) {
@@ -82,63 +75,14 @@ void Synth::process(float* out, uint32_t nFrames, uint32_t sampleRate) {
// updates internal buffered parameters for smoothing
for(auto& p : params_) p.update(); // TODO: profile this
// process all envelopes
// should be easy enough if all the envelopes are in an array to loop over them
gainEnvelope_.set(getParam(ParamId::Osc1VolumeEnvA), getParam(ParamId::Osc1VolumeEnvD), getParam(ParamId::Osc1VolumeEnvS), getParam(ParamId::Osc1VolumeEnvR));
cutoffEnvelope_.set(getParam(ParamId::FilterCutoffEnvA), getParam(ParamId::FilterCutoffEnvD), getParam(ParamId::FilterCutoffEnvS), getParam(ParamId::FilterCutoffEnvR));
resonanceEnvelope_.set(getParam(ParamId::FilterResonanceEnvA), getParam(ParamId::FilterResonanceEnvD), getParam(ParamId::FilterResonanceEnvS), getParam(ParamId::FilterResonanceEnvR));
float gainEnv = gainEnvelope_.process();
float cutoffEnv = cutoffEnvelope_.process();
float resonanceEnv = resonanceEnvelope_.process();
// TODO: envelope is shared between all notes so this sequence involves a note change but only one envelope attack:
// NOTE_A_ON > NOTE_B_ON > NOTE_A_OFF and note B starts playing part-way through note A's envelope
// skip if no active notes
if(!gainEnvelope_.isActive()) {
out[2*i] = 0.0f;
out[2*i+1] = 0.0f;
scope_->push(0.0f);
continue;
// TODO: should I have a write() function ?
// maybe we change the synth.process into just returning a single float and the write can be in audioEngine
}
// assemble float array of parameters so that its easier for voices to retrieve
// TODO: make pitchOffset variable for each oscillator (maybe three values like octave, semitone offset, and pitch offset in cents)
float pitchOffset = 1.0f;
float phaseInc = pitchOffset * 2.0f * M_PI * frequency_ / static_cast<float>(sampleRate);
float gain = gainEnv * getParam(ParamId::Osc1VolumeDepth);
// sample generation
// TODO: wavetables
// TODO: wavetables should be scaled by their RMS for equal loudness (prelim standard = 0.707)
float sineSample = std::sin(phase_);
float squareSample = (phase_ >= M_PI) ? 0.707f : -0.707f;
float sawSample = ((phase_ / M_PI) - 1.0f) / 0.577f * 0.707f;
// switch statement will be replaced with an array index for our array of wavetables
switch (static_cast<int32_t>(std::round(getParam(ParamId::Osc1WaveSelector1)))) {
case 0:
sampleOut = sineSample * gain;
break;
case 1:
sampleOut = squareSample * gain;
break;
case 2:
sampleOut = sawSample * gain;
break;
case 3:
// TODO: no triable wave yet :(
sampleOut = sineSample * gain;
break;
default: // unreachable
break;
// foreach voice, process...
float mix = 0.0f;
for(Voice& v : voices_) {
mix += v.process(&params_[0].current, triggered);
}
// filter sample
float cutoffFreq = cutoffEnv * pow(2.0f, getParam(ParamId::FilterCutoffDepth)) * frequency_;
filter_.setParams(Filter::Type::BiquadLowpass, cutoffFreq, resonanceEnv * getParam(ParamId::FilterResonanceDepth));
sampleOut = filter_.biquadProcess(sampleOut);
// write to buffer
out[2*i] = sampleOut; // left
out[2*i+1] = sampleOut; // right
@@ -148,8 +92,9 @@ void Synth::process(float* out, uint32_t nFrames, uint32_t sampleRate) {
scope_->push(sampleOut); // visualization tap
}
// sampling business
phase_ += phaseInc;
// triggering business
// TODO: get trigger info from voice (lowest frequency voice)
float phase_ = 0.0f;
if (phase_ > 2.0f * M_PI) {
phase_ -= 2.0f * M_PI;
if(!triggered) {