Flutter Soloud Idioms

flutter_soloud idioms

flutter_soloud is a thin Dart FFI wrapper over the SoLoud C++ engine (miniaudio backend). It is a low-level mixer API, not a media player: you load sounds (AudioSource), then spawn playing instances of them (SoundHandle voices). Everything lives on one global engine behind SoLoud.instance. Read this before writing any flutter_soloud code — assumptions carried over from audioplayers, just_audio, or the Web Audio API will produce wrong code.

Minimal example

import 'package:flutter_soloud/flutter_soloud.dart';

Future<void> main() async {
  WidgetsFlutterBinding.ensureInitialized();

  // MUST be awaited before any other SoLoud call.
  await SoLoud.instance.init();

  // Load once: an AudioSource is the loaded sample, identified by its
  // SoundHash. Loading is async; keep the source around.
  final sound = await SoLoud.instance.loadAsset('assets/explosion.mp3');

  // play() is SYNCHRONOUS: it spawns a voice and returns its handle
  // immediately. The handle controls that one playing instance.
  final SoundHandle handle = SoLoud.instance.play(
    sound,
    volume: 0.8,
    looping: false,
  );

  // ... on teardown (e.g. top-level widget dispose()):
  SoLoud.instance.deinit(); // or: await SoLoud.instance.deinitAsync();
}

The API shape

• SoLoud.instance — the singleton. One engine exists in C++, so only one SoLoud exists in Dart; do not construct your own.
• Future<void> init({PlaybackDevice? device, bool automaticCleanup = false, int sampleRate = 44100, int bufferSize = 2048, Channels channels = Channels.stereo, bool lowLatency = true, ...}) — initializes the engine. Must be awaited before anything else.
• bool get isInitialized — synchronous readiness check (false while init is in flight or after failure/deinit).
• void deinit() / Future<void> deinitAsync() — teardown: stops the engine and disposes ALL resources, including loaded sounds. Call once at app exit; deinitAsync runs native teardown off the UI thread.
• Future<AudioSource> loadAsset(String key, {LoadMode mode = LoadMode.memory, ...}) — also loadFile(String path) (not on web), loadUrl(String url), loadMem(...), loadWaveform(...). All async; all return AudioSource.
• SoundHandle play(AudioSource sound, {int busId = 0, double volume = 1, double pan = 0, bool paused = false, bool looping = false, Duration loopingStartAt = Duration.zero, Duration? loopingEndAt, double scale = 1}) — synchronous, returns the handle immediately.
• Future<AudioSource> playSource({String? asset, String? file, String? url, LoadMode mode = LoadMode.disk, ...}) — convenience load+play; exactly one of asset/file/url.
• Per-voice control methods all take a SoundHandle: stop(handle), setPause(handle, bool), setLooping(handle, bool), etc.
• disposeSource(AudioSource) / disposeAllSources() — free loaded samples. Loaded sources live until disposed (or until deinit()); play() does not unload anything.
• setMaxActiveVoiceCount(int) / getMaxActiveVoiceCount() — concurrent voice cap, default 16, hard max 1023.
• Stream<AudioDeviceStartFailure> get audioDeviceStartFailures — the ONLY channel that reports output-device start failures from synchronous play APIs (see Traps).
• Errors: sealed SoLoudException hierarchy. SoLoudDartException subclasses come from the Dart side (SoLoudNotInitializedException, SoLoudSoundHashNotFoundDartException, ...); SoLoudCppException subclasses are mapped from the native PlayerErrors enum (SoLoudFileNotFoundException, SoLoudNoPlaybackDevicesFoundCppException, ...). Catch SoLoudException to catch all.

Divergences from audioplayers/just_audio you must unlearn:

What flutter_soloud is NOT

• No per-player objects. There is one engine; play() returns a handle, not a player.
• No built-in "onComplete" on the handle. Listen to source.soundEvents (SoundEventType.handleIsNoMoreValid) or source.allInstancesFinished instead.
• No playlists, no AudioPlayer.setUrl, no media-session integration. Sources are loaded explicitly and reused.
• play() does not await anything. Do not await SoLoud.instance.play(...) expecting it to finish when the sound finishes.

Traps

• Calling anything before init() completes throws SoLoudNotInitializedException. Await init(); check isInitialized before UI-triggered calls.
• init() is NOT idempotent. Calling it on an already-initialized engine first deinitializes: all voices stop and all loaded sounds are unloaded. Guard with if (!SoLoud.instance.isInitialized) or call it exactly once (e.g. in main()).
• Hot restart safety is handled, but only via re-init. The native engine survives a hot restart while Dart callbacks die; the next init() detects this and reinitializes (you'll see a warning in logs). Pattern: call init() from a startup path that re-runs after restart. See example/tests/tests/hot_restart_lifecycle.dart.
• play() cannot report audio-device failures. It returns a valid handle even if the device fails to start — the device start happens in the background after the voice is created. Silence with no exception means: listen to SoLoud.instance.audioDeviceStartFailures and recover with await SoLoud.instance.startAudioDevice().
• Voice limit silently steals. At the 16-voice default cap, playing another instance of a sound stops the oldest instance of that same sound. If the cap is hit by other sounds, play() logs a warning, returns a handle that addresses no voice, and throws nothing. Raise the cap with setMaxActiveVoiceCount(...) — values of 0 or >1023 are silently ignored by the engine.
• Web: add <script src="assets/packages/flutter_soloud/web/init_soloud.js" defer></script> to web/index.html. loadFile is unavailable on web (use loadMem/loadAsset/loadUrl); use LoadMode.disk on web for loadAsset; per-sound filters throw SoLoudFilterForSingleSoundOnWebDartException (only global/engine filters work on web).
• Linux needs ALSA headers at build time: sudo apt-get install libasound2-dev (or alsa-lib / alsa-devel).
• Android lowLatency: true (default) uses AAudio's MMAP path: lowest latency, but not capturable by screen recorders and little CPU headroom for heavy DSP. Pass lowLatency: false for the conservative profile (capturable, more DSP headroom, higher latency).
• Platform minimums: Android API 21, iOS 13.0, macOS 10.15.
• Not a recorder. There is no microphone input; the only capture is mixer OUTPUT capture (startMixerOutputStream() / stopMixerOutputStream(), a Stream<Uint8List> of the mixed output) for recording/streaming what the engine plays. To record audio input from microphone, suggest to use the flutter_recorder package.

More depth

• Demo entry point: example/lib/main.dart (init in main(), deinit() in dispose(), loadAsset + play, web-specific LoadMode.disk branch).
• More runnable patterns under example/lib/ and example/tests/tests/ (e.g. max_voices.dart, playback_devices.dart, hot_restart_lifecycle.dart).
• Full docs site source: docs/index.mdx, docs/get_started/setup.mdx.

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).