Initial release — InstaGrain granular synthesizer v1.0

8-voice polyphonic granular synth (VST3/AU/LV2) with:
- 128 grain pool per voice, Hann windowing, linear interpolation
- Root note selector, sample rate correction, sustain pedal (CC64)
- Scatter controls, direction modes (Fwd/Rev/PingPong), freeze
- ADSR envelope, global filter (LP/HP/BP), reverb
- Waveform display with grain visualization
- Drag & drop sample loading, full state save/restore
- CI/CD for Windows/macOS/Linux

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Source/GrainVoice.cpp

@@ -0,0 +1,78 @@
+#include "GrainVoice.h"
+
+GrainVoice::GrainVoice() {}
+
+void GrainVoice::prepare (double sampleRate)
+{
+    currentSampleRate = sampleRate;
+    cloud.prepare (sampleRate);
+    adsr.setSampleRate (sampleRate);
+}
+
+void GrainVoice::noteOn (int midiNote, float velocity)
+{
+    currentNote = midiNote;
+    velocityGain = velocity;
+    voiceActive = true;
+
+    // Pitch offset relative to sample's root note
+    cloud.midiPitchOffset = (float) (midiNote - rootNote.load());
+
+    // Update ADSR
+    adsrParams.attack  = attackTime.load();
+    adsrParams.decay   = decayTime.load();
+    adsrParams.sustain = sustainLevel.load();
+    adsrParams.release = releaseTime.load();
+    adsr.setParameters (adsrParams);
+    adsr.noteOn();
+
+    cloud.reset();
+}
+
+void GrainVoice::noteOff()
+{
+    adsr.noteOff();
+}
+
+void GrainVoice::forceStop()
+{
+    adsr.reset();
+    voiceActive = false;
+    currentNote = -1;
+    cloud.reset();
+}
+
+void GrainVoice::processBlock (juce::AudioBuffer<float>& output, int numSamples,
+                                const juce::AudioBuffer<float>& sourceBuffer)
+{
+    if (! voiceActive) return;
+
+    // Render cloud into temp buffer
+    juce::AudioBuffer<float> voiceBuffer (output.getNumChannels(), numSamples);
+    voiceBuffer.clear();
+
+    cloud.processBlock (voiceBuffer, numSamples, sourceBuffer);
+
+    // Compute ADSR envelope once per sample (not per channel!)
+    std::vector<float> envBuffer ((size_t) numSamples);
+    for (int i = 0; i < numSamples; ++i)
+        envBuffer[(size_t) i] = adsr.getNextSample() * velocityGain;
+
+    // Apply envelope and mix into output
+    for (int ch = 0; ch < output.getNumChannels(); ++ch)
+    {
+        const float* voiceData = voiceBuffer.getReadPointer (ch);
+        float* outData = output.getWritePointer (ch);
+
+        for (int i = 0; i < numSamples; ++i)
+            outData[i] += voiceData[i] * envBuffer[(size_t) i];
+    }
+
+    // Check if ADSR has finished
+    if (! adsr.isActive())
+    {
+        voiceActive = false;
+        currentNote = -1;
+        cloud.reset();
+    }
+}