Independent character that returns.
Pitch, cutoff, envelope timing and amplifier behaviour vary by voice. The variation and slow drift remain reproducible, so the width belongs to the patch and returns with the session.
OCTET Specialist series
WORKING ENGINE + EDITOR · CALIBRATION OPEN
Eight complete voices. One instrument.
Octet is modelled on the Oberheim SEM voice card used in the Four Voice and Eight Voice, with a two-pole state-variable filter that moves continuously from low-pass through notch to high-pass and offers band-pass separately.
It is not an OB-Xa or OB-8 study. Those later instruments used CEM3320 filters with two-pole and four-pole modes, while Octet stays with the earlier state-variable architecture.
Instrument panelOctet
01 The instrument
A complete circuit for every note.
Eight related voices make chords move without becoming arbitrary. Each voice carries the full oscillator, filter, envelope, modulation and amplifier path, with stable component variation derived from the patch seed.
One control crosses four shapes.
The mode path moves from low-pass through a deep central notch to high-pass. Band-pass remains a separate selection, and resonance rises without structural bass loss.
From hollow leads to glassy pads.
Hard sync, pulse-width movement, vocal filter sweeps, stable unison groups and repeatable voice width give a direct route to animated leads, open chords and shifting pads.
02 Working build
The instrument, without guesswork.
The list separates what the engine and editor do now from format, preset and calibration work that remains open.
- Voices
- Eight-voice default with a prepared pool selectable from one to sixteen voices
- Oscillators
- Two discrete saw-core oscillators per voice, saw and variable-width pulse, oscillator 2 hard sync and pulse-width modulation
- Filter
- Two-pole OTA state-variable filter, continuous low-pass through notch to high-pass mode, separate band-pass and resonance
- Envelopes
- Two resistor-capacitor ADS envelopes with release following the decay timing
- Modulation
- Per-voice LFO with sine, square and pulse shapes, plus sample-and-hold clocked from seeded noise
- Playing modes
- Poly, unison and configurable poly-unison groups, round-robin allocation, quiet-release voice stealing and glide
- Output
- Stereo main output with program-level pan and repeatable per-voice character
- Current format
- Working CLAP instrument. VST3 and Audio Unit remain later format work; release system requirements are not yet established
- Program state
- The current editor uses one shared program across the voice pool and starts from Init. A categorized factory bank and per-voice programmer are not part of the current build
- Current boundaries
- No sub oscillator, effects, arpeggiator, sequencer, microtuning, modulation matrix or MPE
03 Evidence
Results and limits travel together.
See the oscillator targets, filter-equation conformance, rate checks, repeatable polyphony and the calibration and performance work that remains open.