Flutter Soloud Playback

flutter_soloud is not a "player object" API like audioplayers or just_audio. You load audio once into an AudioSource (the sample data, kept in RAM or on disk), then start any number of playing instances of it with play(). Each instance is a SoundHandle — an ephemeral voice that becomes invalid the moment it stops or ends. All control methods (setPause, seek, setVolume, ...) take a SoundHandle, not the source.

Minimal example

import 'package:flutter_soloud/flutter_soloud.dart';

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

  final source = await SoLoud.instance.loadAsset('assets/music.mp3');

  // play() is synchronous and returns immediately.
  final handle = SoLoud.instance.play(
    source,
    volume: 0.8,
    pan: 0.0, // -1.0 left .. 0.0 center .. 1.0 right
  );

  SoLoud.instance.seek(handle, const Duration(seconds: 10));
  SoLoud.instance.setPause(handle, true);   // pause
  SoLoud.instance.setPause(handle, false);  // resume
  await SoLoud.instance.stop(handle);       // completes when the voice really ends

  await SoLoud.instance.disposeSource(source);
}

The API shape

All methods below live on the singleton SoLoud.instance and throw SoLoudNotInitializedException if called before await SoLoud.instance.init() completes.

Starting playback

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,
  int? loopingStartOffsetAt,
  int? loopingEndOffsetAt,
  double scale = 1,
})

• Synchronous and non-blocking — do not await it (unlike just_audio's play()).
• The sound starts at the beginning of the next output buffer (~0–46 ms at the default 2048-frame buffer / 44.1 kHz). The AudioSource stays loaded and replayable until disposeSource().
• scale is the initial speed multiplier (1.0 = normal); available on all play* variants.
• Max concurrent voices defaults to 16 (setMaxActiveVoiceCount).

Future<AudioSource> playSource({
  String? asset, String? file, String? url,  // exactly one
  LoadMode mode = LoadMode.disk,
  int busId = 0, double volume = 1, double pan = 0, bool paused = false,
  bool looping = false,
  Duration loopingStartAt = Duration.zero, Duration? loopingEndAt,
})

• Fire-and-forget load+play. Loads with autoDispose: true, so the source frees itself when playback ends.
• Returns the AudioSource, not a SoundHandle — you cannot control the voice afterwards. If you need pause/seek/stop, call loadAsset() + play() yourself.
• There is a load lag between the call and the audible start; don't use it for latency-sensitive one-shots.

Pause / resume

SoLoud.instance.pauseSwitch(handle);        // toggle
SoLoud.instance.setPause(handle, true);     // pause
SoLoud.instance.setPause(handle, false);    // resume
final isPaused = SoLoud.instance.getPause(handle);

Stopping

await SoLoud.instance.stop(handle);   // Future completes when the voice has actually ended
SoLoud.instance.stopAll();            // every voice, synchronous
SoLoud.instance.stopAudioSource(source);  // every voice playing that source

• stop() awaits the native voice-ended event (with an internal ~300 ms fallback). Stopping an already-ended handle is a no-op, not an error.
• None of these dispose the source — it stays loaded and replayable. Use disposeSource(source) to free memory.
• Do not confuse with stopAudioDevice(), which suspends the output device but leaves voices untouched.

Position and seek

SoLoud.instance.seek(handle, const Duration(seconds: 30));
final pos = SoLoud.instance.getPosition(handle);      // play head of this voice
final len = SoLoud.instance.getLength(source);        // total length of the source
final t = SoLoud.instance.getStreamTime(handle);      // how long this voice has been playing

• getLength takes the AudioSource, not a handle. getPosition/getStreamTime take handles.
• Negative seek times are clamped to Duration.zero.
• For BufferingType.released streams, getPosition is meaningless — use getStreamTimeConsumed(source) instead.

Looping

Loop regions are half-open: [start, end). A null end means the source's natural end.

// At play time (preferred — applies from the first decoded sample):
final h = SoLoud.instance.play(
  source,
  looping: true,
  loopingStartAt: const Duration(seconds: 1),
  loopingEndAt: const Duration(seconds: 5), // exclusive
);
// Or frame offsets (mutually exclusive with the Duration params):
final h2 = SoLoud.instance.play(
  source,
  looping: true,
  loopingStartOffsetAt: 44100,
  loopingEndOffsetAt: 220500,
);

// On a live voice:
SoLoud.instance.setLooping(h, true);
SoLoud.instance.setLoopPoint(h, const Duration(seconds: 1));   // start
SoLoud.instance.setLoopEndPoint(h, const Duration(seconds: 5)); // null = natural end
final looping = SoLoud.instance.getLooping(h);
final start = SoLoud.instance.getLoopPoint(h);
Duration? end = SoLoud.instance.getLoopEndPoint(h);

Live setLoopPoint/setLoopEndPoint changes are reflected in the getters immediately but only apply to playback at the next source refill — up to ~512 already-decoded frames plus backend latency still use the old region. Pass the bounds to play() when they must hold from sample one.

Playback speed

final h = SoLoud.instance.play(source, scale: 1.5); // initial speed
SoLoud.instance.setRelativePlaySpeed(h, 2.0);        // live change
final speed = SoLoud.instance.getRelativePlaySpeed(h);

Speed changes the effective sample rate (pitch shifts with it — there is no built-in time-stretch). Faster = more CPU and memory; slower is cheaper.

Voice introspection

final n = SoLoud.instance.getActiveVoiceCount();  // currently audible voices
final m = SoLoud.instance.getVoiceCount();        // voices the app asked to play
final k = SoLoud.instance.countAudioSource(source); // voices playing this source
final ok = SoLoud.instance.getIsValidVoiceHandle(handle); // still playing/paused?
SoLoud.instance.setMaxActiveVoiceCount(32); // default 16, hard max 1023

Protection and inaudible behavior

SoLoud.instance.setProtectVoice(musicHandle, true); // don't kill this when the voice limit hits
SoLoud.instance.setInaudibleBehavior(handle, false, true); // (mustTick, kill); for 3D voices

Voice groups

final group = SoLoud.instance.createVoiceGroup(); // returns a SoundHandle, throws on failure
SoLoud.instance.addVoicesToGroup(group, [handle1, handle2]);
SoLoud.instance.setPause(group, true);            // group handle works wherever a voice handle does
SoLoud.instance.isVoiceGroup(group);              // true for group handles
SoLoud.instance.isVoiceGroupEmpty(group);         // ended voices are trimmed automatically
SoLoud.instance.destroyVoiceGroup(group);         // does NOT stop the voices in it

Traps

• Handles die silently. A SoundHandle becomes invalid when the voice ends, is stopped, or is killed to make room at the voice limit. Most control calls on a dead handle either do nothing or throw SoLoudSoundHandleNotFoundCppException (setPause, pauseSwitch). Guard with getIsValidVoiceHandle() when in doubt.
• Voice limit kills quietly. At the max active voice count (default 16), playing a sound whose source already has voices kills the oldest one; otherwise the new sound simply does not play — no exception, and the returned handle addresses nothing. Protect music with setProtectVoice and/or raise setMaxActiveVoiceCount (values of 0 or >1023 are silently ignored).
• setRelativePlaySpeed(handle, 0) is undefined behavior, likely a crash. The lower clamp is 0.05, applied silently.
• Mixing Duration-based and frame-offset looping params (loopingStartAt/loopingEndAt vs loopingStartOffsetAt/loopingEndOffsetAt) is mutually exclusive and fails an assert in debug builds; invalid offsets throw ArgumentError.
• MP3 + LoadMode.disk seek lag. Seeking a disk-loaded MP3 is slow (the codec must walk frames to find the position). For a music-player-style seek bar, load with LoadMode.memory or use a seekable format.
• Released buffer streams can't seek. seek() on a stream created with BufferingType.released throws SoLoudBufferStreamWithReleasedBufferTypeCannotBeSeekedCppException; an unseekable handle throws SoLoudInvalidParameterException.
• stopAll()/stopAudioSource() do not free memory. Sources stay loaded; forgetting disposeSource() is the common leak.
• play() latency is up to one output buffer. Fine for reactive one-shots; wrong for rhythmic/scheduled playback — that is the sibling skill's territory.
• Calling any method before await SoLoud.instance.init() finishes throws SoLoudNotInitializedException (or the call is ignored). Re-calling init() deinitializes first: all voices stop and all sources unload.

More depth

• Sample-accurate timing (playClocked, playScheduled, setDelaySamples, getEngineTime): see the sibling skill flutter-soloud-scheduling.
• Working playback demo with play/pause/seek/loop UI: example/lib/audio_context/audio_context.dart in the plugin repo.
• Full doc page: docs/audio/playback.mdx (published at docs.page/alnitak/flutter_soloud).

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).