@maruaudio/dsp/envelopes/adsr is an exponential ADSR state machine in VCA form, the same one the subtractive voice runs. Each stage moves the level one f32 step toward its target, level += (target − level)·c, with c = 1 − exp(−4.6 / (t·rate)) so a stage gets about 99 % of the way in its time (@maru/voices-stages). Attack ends at 0.999 and jumps to 1, decay ends within 0.001 of sustain, release ends at 0.0005 and goes idle. Gate-on is a full retrigger from level 0; gate-off releases from the current level.

const env = initAdsrNode(48000)
setAdsrNodeRelease(env, 0.8)
adsrNodeGateOn(env)
const gain = adsrNodeTick(env) // per sample

Parameters

Times below 0.0001 s are floored to 0.0001 s in the coefficient.

Lifecycle

initAdsrNode(sampleRate) → setters → adsrNodeGateOn and adsrNodeGateOff → adsrNodeTick per sample → resetAdsrNode, which returns to idle at level 0 and keeps the times. Gate-off while idle does nothing. Stages are ADSR_STAGE_IDLE (0) to ADSR_STAGE_RELEASE (4).

When to use it

Shaping the amplitude or filter of a synth voice, with an exponential curve and a retrigger that always restarts the attack from silence.

Limits

• One envelope, one sample per tick; the level is f32.
• The coefficient is recomputed from the stored time on every attack, decay and release sample, so each such tick calls exp.
• Gate-on restarts from 0, which can click on a sounding note; there is no legato mode.
• audio-kernels keeps the PCM16 render entry.