Flutter Soloud 3d Audio

flutter_soloud wraps the SoLoud C++ engine's built-in 3D audio. Every voice started with play3d has a 3D position and velocity; a single global "listener" (usually your camera or player) hears them. The coordinate system is right-handed (Y up, X right, Z toward the viewer). Unlike audioplayers/just_audio there is no asset "player" object per sound: you get a SoundHandle per playing instance and mutate its 3D state through SoLoud.instance setters. Also unlike the web Audio API, SoLoud does not move anything for you — positions and velocities are snapshots you must update yourself each tick.

Minimal example

import 'package:flutter_soloud/flutter_soloud.dart';

Future<void> main() async {
  await SoLoud.instance.init();

  final sound = await SoLoud.instance.loadAsset('assets/sfx/whoosh.mp3');

  // Listener at origin, looking down -Z, head up along +Y.
  SoLoud.instance.set3dListenerParameters(
    0, 0, 0, // position
    0, 0, -1, // at (looking direction)
    0, 1, 0, // up
    0, 0, 0, // velocity (needed for Doppler)
  );

  // Play at (10, 0, -5) moving toward the listener at 5 units/s.
  final handle = SoLoud.instance.play3d(
    sound,
    10, 0, -5, // posX, posY, posZ (positional!)
    velX: -5, // velocity, for Doppler
    looping: true,
  );

  // Required to hear distance falloff — default attenuation is NONE.
  SoLoud.instance.set3dSourceAttenuation(handle, 2, 1); // linear
  SoLoud.instance.set3dSourceMinMaxDistance(handle, 1, 50);
}

The API shape

All methods are on SoLoud.instance (singleton), all positions/velocities are double in your world units. Playback methods return SoundHandle synchronously (not a Future) — different from the Future<...>-returning patterns in other packages.

Playback (all throw SoLoudNotInitializedException if you skipped init):

• SoundHandle play3d(AudioSource sound, double posX, double posY, double posZ, {double velX = 0, double velY = 0, double velZ = 0, int busId = 0, double volume = 1, bool paused = false, bool looping = false, Duration loopingStartAt = Duration.zero, Duration? loopingEndAt, int? loopingStartOffsetAt, int? loopingEndOffsetAt, double scale = 1}) — one-shot, lowest latency; start is quantized to output-buffer boundaries.
• SoundHandle play3dClocked(AudioSource sound, Duration soundTime, double posX, double posY, double posZ, {...}) — pass your app's monotonically increasing "physics time"; the engine spaces rapid-fire sounds sample-accurately so they don't clump at the next buffer boundary. Costs ~2 buffers of latency. Use for machine-gun / rhythmic 3D SFX.
• SoundHandle play3dScheduled(AudioSource sound, Duration atTime, double posX, double posY, double posZ, {Duration? duration, ...}) — starts at an absolute engine time (SoLoud.instance.getEngineTime()), sample-accurate; duration auto-stops it at atTime + duration.

Listener (one global listener; default position is (0,0,0)):

• set3dListenerParameters(posX, posY, posZ, atX, atY, atZ, upX, upY, upZ, velocityX, velocityY, velocityZ) — 12 positional doubles, no named parameters.
• set3dListenerPosition(x, y, z) / set3dListenerAt(x, y, z) / set3dListenerUp(x, y, z) / set3dListenerVelocity(x, y, z) — individual setters. at is the look direction; up is the head-up vector.

Per-source (take the SoundHandle returned by play3d*):

• set3dSourceParameters(handle, posX, posY, posZ, velocityX, velocityY, velocityZ) — positional doubles.
• set3dSourcePosition(handle, x, y, z) / set3dSourceVelocity(handle, x, y, z).
• set3dSourceMinMaxDistance(handle, minDistance, maxDistance) — defaults are 1 and 1000000. Inside min the sound is at full volume; beyond max it no longer gets quieter.
• set3dSourceAttenuation(handle, int attenuationModel, double attenuationRolloffFactor) — models: 0 NO_ATTENUATION, 1 INVERSE_DISTANCE, 2 LINEAR_DISTANCE, 3 EXPONENTIAL_DISTANCE. Default is 0 (none) with rolloff 1.
• set3dSourceDopplerFactor(handle, double dopplerFactor) — 0 disables, 1 normal, >1 exaggerated.

Global Doppler tuning:

• set3dSoundSpeed(double speed) / double get3dSoundSpeed() — default 343 (meters/second in dry air at ~20 °C). SoLoud doesn't know your unit scale, so if 1 world unit ≠ 1 meter, scale the sound speed accordingly or Doppler will be wrong.

Per-frame update pattern

Nothing moves unless you move it. In your game loop / Ticker / onTick:

void update(double dt) {
  // Listener follows the camera/player every frame.
  SoLoud.instance.set3dListenerParameters(
    cam.x, cam.y, cam.z,
    cam.forwardX, cam.forwardY, cam.forwardZ,
    cam.upX, cam.upY, cam.upZ,
    cam.velX, cam.velY, cam.velZ, // for Doppler
  );

  // Each moving emitter updates position AND velocity.
  for (final e in movingEmitters) {
    SoLoud.instance.set3dSourceParameters(
      e.handle, e.x, e.y, e.z, e.vx, e.vy, e.vz,
    );
  }
}

Traps

• Default attenuation is 0 (NO_ATTENUATION). If you play3d and never call set3dSourceAttenuation, you get left/right panning from position but no volume falloff with distance — the #1 "3D doesn't work" report. Pick a model (linear 2 is a good default for games) and set sensible min/max distances (e.g. 1–50 for meter-scale worlds; leaving max at the 1000000 default makes falloff imperceptible).
• Positions are snapshots. A sound stays where play3d put it until you call set3dSourcePosition/set3dSourceParameters. No automatic binding to a game object; update every tick while the voice is alive.
• Velocity doesn't move the source; it only feeds Doppler. You must supply both the position delta (via position updates) and the velocity vector yourself. Same for the listener. Zero velocity = no Doppler.
• at and up must not be parallel (and should be roughly orthogonal and normalized). If both are e.g. (0,1,0) the listener orientation is degenerate and panning/orientation glitches or collapses.
• set3dListenerParameters takes 12 positional doubles — no named arguments, and the source setters are positional too. Don't pass named args copied from the web docs; it won't compile.
• play3d returns SoundHandle synchronously. No await. The async part is init() and loadAsset/loadFile.
• Handle lifetime: once a non-looping sound finishes, its handle is invalid; per-source setters on a dead handle silently do nothing. Track sound.soundEvents / allInstancesFinished if you reuse handles.
• Voice limit: default max active voices is 16 (setMaxActiveVoiceCount to change). At the limit, the oldest instance of the same sound is stolen; if none of that sound is playing, the new play silently produces a handle addressing no voice.
• play3d start is buffer-quantized. Rapidly triggered SFX (gunfire) clump to buffer boundaries — use play3dClocked (or play3dScheduled with getEngineTime()) for sample-accurate spacing.
• Doppler scale mismatch: set3dSoundSpeed default 343 assumes meters. If your world is in pixels or generic units, audible Doppler warble at game speeds means your velocities are huge relative to 343 — raise the sound speed or scale down.
• Unit consistency: attenuation distances, listener/source positions, and sound speed all share the same implicit unit. Mixing meters for distances and pixels for positions gives nonsense falloff.

More depth

• Working test exercising the full API (listener, source, attenuation, Doppler, sound speed): example/tests/tests/three_d_audio.dart. The example app (example/lib/) has no dedicated 3D demo screen.
• Upstream SoLoud 3D concepts (attenuation models, Doppler math): https://solhsa.com/soloud/concepts3d.html

Keeping this skill current

This skill ships inside the flutter_soloud package, so upgrading flutter_soloud can carry a newer revision of it than the copy installed in the project. To check, run:

dart run flutter_soloud:skills --check

It reports the installed and bundled skill versions and exits non-zero when an update is available. Offer to update with dart run flutter_soloud:skills (which touches only the skills, never pubspec or build files).