Ambi Encoder Horizon

Ambi Encoder Horizon is an autonomous, zero-input instrument for the audible edge of an outdoor sound field: a bell carrying across a valley, machinery behind a ridge, traffic beyond fields, aircraft on a distant approach, foghorns past a headland, or surf behind dunes. Up to 32 independent synthetic entities form a selectable first- through seventh-order ACN/SN3D scene.

s3g Ambi Encoder Horizon has no audio input and exposes one fixed 64-channel output bus.

Workflow

1. Set the track to 64 channels, insert s3g Ambi Encoder Horizon, and choose 1OA through 7OA from ORDER.
2. Follow it with an ambisonic decoder.
3. Start from a factory scene and let it run. Horizon is autonomous; it needs no track audio, MIDI, or host tempo.
4. Choose an ECOLOGY, then balance its population with ENTITIES, ACTIVITY, OCCUPANCY, and PACE.
5. Use SIGNALS, LANDSCAPE, and LOCAL FLOOR to separate the three audible layers before changing distance or atmosphere.
6. Set RANGE, ARC, and DETAIL, then adjust air, ground, terrain, carry, and turbulence. Add the Field Listener last if the scene should react to what it has recently produced.

Signal Flow

1. The ecology and deterministic SEED select as many as 32 entities from flow, rolling body, machinery, modal bells, air, traffic, aircraft, foghorn, and surf engines.
2. The environmental score gives each entity finite activity and rest arcs. ACTIVITY, OCCUPANCY, PACE, and MEMORY shape when events emerge and how long the scene remembers them.
3. Every mono entity receives its own distance, air-loss, ground, terrain, carry, and turbulence path before encoding.
4. Non-local sources also feed a sparse four-direction outdoor return representing separated ground, shoreline, ridge, or irregular-topography paths rather than an indoor room.
5. Direct entities and outdoor returns are encoded into the selected 1OA–7OA field. Lower-coherence returns can spread while a bell, horn, vehicle, or aircraft remains locatable.
6. The optional Field Listener analyzes the completed pre-output field and applies a bounded perceptual response. OUT remains outside the listening loop.

Three Scene Layers

The local and landscape layers are synthesized environmental bodies, not recorder noise. AIR NOISE is a separate source control and can reach true zero without disabling event motion, ground response, outdoor returns, or ambisonic detail.

Acoustic-Horizon View

The field view shows the scene’s current entities, selected ecology and ground, perceptual range, and output format. TOP and SIDE provide fixed projections, while 3/4 restores perspective. Drag empty field space to orbit and use − or + to zoom. Camera changes affect only the view and are stored with plugin state.

RANGE spans a perceptual 0.03–20 km scale rather than predicting calibrated sound pressure. AZIMUTH and ELEVATION place the scene, ARC sets its angular extent, DETAIL changes spatial coherence, and EDGE sets the horizon-signal balance. Restrained geometric loss and CARRY preserve the faint audibility that a strict inverse-square model would remove.

Ecologies and Generators

The ecology choices are MIXED, RURAL, TRAFFIC, CITY, INDUSTRIAL, WATER, WEATHER, AIRPORT, and COAST. Ecology selects which source families and model controls are available; each family level independently reaches silence.

The generator panel changes with the ecology. Airport scenes expose aircraft amount, continuous FLIGHT state, pass speed, engine power and tone, plus related traffic controls. Water and Coast expose foghorn amount, horn pitch, pressure and call length, together with surf, wave rate, and wave break. Other ecologies show the machine, bell, traffic, aircraft, or surf controls relevant to that environment.

FLIGHT moves continuously from landed and taxi behavior through takeoff or approach to an overhead pass, jointly changing path width, elevation, speed, load, closest approach, and restrained Doppler. Foghorn pressure changes drive, upper partials, envelope, and air-column response; surf is built from overlapping asymmetric wave cycles rather than a static noise bed.

Environmental Score

MEMORY controls continuity between recent events, while CASCADE encourages delayed spatial relationships. SIGNALS, LANDSCAPE, and LOCAL FLOOR set the three-layer balance. Generator events use distinct target and audible envelopes, with slower attack and release at greater perceptual range so a far source emerges instead of arriving as a close click.

SEED defines a repeatable scene identity. The same seed and state reproduce the same deterministic structure, while the score continues to evolve rather than looping a loaded recording.

Atmosphere and Outdoor Returns

AIR LOSS removes high-frequency energy with distance. GROUND chooses Water, Hard, Mixed, Grass, or Forest behavior. TERRAIN increases obstruction and return character; CARRY permits stronger favorable propagation; and TURBULENCE adds slow and fast amplitude variation. These parameters shape each source path before the final ambisonic mix.

The return network uses separated outdoor paths and a dark, low-feedback scattered tail. Water and Coast favor longer, lower-elevation reflections; land ecologies use shorter terrain-dependent paths. It supplies environmental depth without pretending to be a calibrated weather, terrain-diffraction, or room-acoustics solver.

Field Listener

The eight-pickup Cube listener follows the shared Listener Mode design. FOLLOW prefers recently audible directions, COUNTER evaluates their antipodes, and BALANCE prefers less exposed regions. OFF is an exact open-loop reference. AMOUNT scales the response without feeding audio back into a generator.

REACH makes small changes to audibility, outdoor-return balance, and apparent range. GLIMPSE uses novelty and short-term charge to reveal faint events without retriggering them. SETTLE restrains persistently exposed directions and moves some energy toward the landscape. DISTANCE combines direct-to-return balance, spectral tilt, and habituation to adjust loss, coherence, direct level, and return together.

Controls

Order and Routing

The output is ACN/SN3D. Place compatible Ambisonic effects between Horizon and the decoder only when their supported order and channel count match the active field.

Factory Scenes and State

The sixteen deterministic scenes are BELL ACROSS VALLEY, CLEAR NIGHT INVERSION, HIGHWAY BEYOND FIELDS, CITY BEYOND RIDGE, INDUSTRIAL NIGHT REACH, ACROSS STILL WATER, DAWN LONG HORIZON, STORM BEYOND HILLS, DISTANT RAIL CORRIDOR, QUIET AGRICULTURAL BASIN, SETTLEMENT THROUGH FOREST, VANISHING ACOUSTIC HORIZON, HIGH FIDELITY HORIZON, AIRPORT APPROACH CORRIDOR, FOGHORNS BEYOND HEADLAND, and OPEN OCEAN BEYOND DUNES.

Factory selection and RANDOM preserve OUT, ORDER, and the current listener mode, amount, and response. LOAD and SAVE exchange complete custom Horizon states. Camera mode, orbit, and zoom are also serialized.

Limits

Horizon is fully synthetic and loads no sound files. Its kilometer display is perceptual, not a measurement of source power, weather, terrain, or listener sound pressure. The propagation and auditory-distance controls are musical approximations; see Outdoor Sound Propagation and Auditory Distance for the acoustic context.