Version History

v3.1.0 (dev) - October 2026

• Dev snapshots for testing in other projects; see CHANGELOG.md at the package root and “Changes since 3.0.0” in BETA_KNOWN_ISSUES.md for what changed.
• Scoped parameters: AudioParameter assets, AudioEmitter, per-play values on AudioHandle, global SetParameter/RampParameter, AudioEvent.Post(AudioEmitter), AudioManager.Post, and AudioManager.UpdateMode/Tick. The string RTPC API is the global scope. Blend Containers read their parameter per play, once a frame; BlendContainer.UpdateBlend is obsolete.
• Parameter curves, regions and pitch follow: ParameterCurve (volume dB, pitch cents, low- and high-pass Hz) on any container and Blend entry; Blend Layout (Volume Curves or Regions), Crossfade shape and AutoLayout; PitchFollowParameter with per-entry rootValue, followPitch and followRangeSemitones; GainStack.CurveGain; per-voice filters for filter curves.
• Nested containers: every play keeps a tree of the containers it reached, so nested Blends read their parameters per play and late voices get their Blend’s curve and Volume. New AudioContainer.UsesParentPositioning / UsesParentOutput.
• Listening points: AudioManager.AddListeningPoint / RemoveListeningPoint / ListeningPoints / DriveUnityListener, AudioListeningPoint, and ListenerUtil.TryGetPosition. With a point registered the scene is never searched; without one, the search warns once and runs at most once a frame.
• Parameter smoothing: ParameterSlew (None, Rate, Ease with separate rise and fall) on AudioParameter, per-emitter and per-play overrides, SnapParameter / SnapAllParameters, and the Bypass Slew While parameter.
• Virtual voices keep their place: positions in samples on the audio clock at the voice’s pitch (AudioVoice.VirtualSamples), a short fade around virtualizing and realizing (GainStack.RealizeGain, GetGainWithoutRealize), and AudioContainer.DopplerLevel.
• The play is the unit of the voice pool: whole-play steals after making room (AudioContainer.EstimateVoiceCount), plays real or virtual as a whole with stickiness, one start time per play, BlendContainer.LoopMode / SilentLayers / SilentThresholdDb (Loop obsolete), AudioEvent.Persistent, Unity voice-limit warnings and AudioSource.priority.
• No forced multiplayer code: the SFXSYSTEM_DISABLE_NETCODE define leaves the netcode assemblies out; PropagationManager hears from the registered listening point.
• Named emitter points: AudioEmitter Points, Post(event, "Exhaust") / Post(event, number), FindPoint, SetPoint, GetPointOffset / GetPointPosition, EmitterPoint; Velocity, SetVelocity, ClearVelocity; AudioEvent.Post(AudioEmitter, point) and AudioManager.Post(event, emitter, point).
• Switches per sound, following while they play: AudioSwitchGroup assets (names still work), with DrivenBy, Ranges and Hysteresis; SetSwitch / ClearSwitch / TryGetSwitch / GetEffectiveSwitch on AudioEmitter and AudioHandle; AudioManager.SetSwitch(AudioSwitchGroup, value), ClearSwitch, GetSwitch(AudioSwitchGroup); SwitchContainer.SwitchGroup, WhenSwitchChanges and CrossfadeTime; GainStack.SwitchGain.
• Distance assets and built-in values: AudioAttenuation (volume, low-pass, high-pass and spread over metres, Max Distance), AudioContainer.Attenuation, ContainerOverrides.attenuation, AudioHandle.SetAttenuation / ClearAttenuation; AudioParameter.ValueFrom (ParameterSource: Distance, ApproachSpeed, Angle) and IsBuiltIn. AudioContainer.MaxDistance’s setter writes a runtime override.
• Audition outside Play mode: Window > Audio System > Audition and AudioAudition (AuditionLayer, AuditionInputs): play a container tree with parameter sliders, switch values, a sweep and a layer readout while editing it.
• One-shots per emitter: AudioEvent.MaxInstancesPerEmitter / CooldownPerEmitter; PostValues and the Post overloads that take it (event, emitter, manager); AudioManager.RandomNumbers (AudioRandomSource: Pepper’s own streams by default, or Unity); AudioEmitter.RandomSeed / SetRandomSeed; Random containers’ Avoid Repeat Last per emitter.
• Bus property curves: AudioBus.ParameterCurves (Volume, Low Pass, High Pass on global values), LowPassParameter / HighPassParameter (exposed mixer cutoffs), CurveVolumeGain, LowPassHz, HighPassHz; the bus inspector’s Add Low Pass / High Pass buttons.
• Lifecycle, teardown and pause (3.1.0-dev.5): AudioManager.PersistAcrossScenes and Pepper’s own “AudioVoices” root (switching the manager off changes nothing); AudioManager.Inert for dedicated servers; AudioEmitter.WhenDisabled (EmitterDisableMode: Keep, Stop, Pause) and StopFadeSeconds; EventAction.scope (ActionScope: All, ThisEmitter); AudioBus.Pause / Resume / IsPaused / WhilePaused (BusPauseBehaviour); AudioEvent.TimeMode and AudioBus.DuckingTimeMode (AudioTimeMode: Scaled, Unscaled). Finished handles ignore new OnStarted / OnStopping / OnLoop subscriptions.
• Tools (3.1.0-dev.6): the Audition plays events too, runs any number of sweeps (SweepMode), replays recordings and shows written values (AudioAudition); the Live View with AudioManager.GetPlaysDebug, AudioPlayDebugInfo, VoiceVirtualReason, the new AudioVoiceDebugInfo fields, AudioHandle’s value listing and AudioEmitter.AllEnabled; AudioEmitterRecorder and AudioParameterRecording; Loop Analysis; the Engine Builder; the Container Validator’s project-wide rules and ValidateEvent; folder import rules (no settings asset ships in the package); AudioStatistics.poolSize; undoable bus faders.
• Routing, mix and positions (3.1.0-dev.7): offsets that stick: AudioEmitter.VolumeOffsetDb / PitchOffsetCents, AudioHandle.SetVolumeOffset / VolumeOffsetDb / VolumeScale / PitchOffset, PostValues.WithVolumeOffset / WithPitchOffset / VolumeOffset, GainStack.EmitterGain, PitchStack.Emitter, and a handle’s SetVolumeScale / SetPitchOffset reaching voices it starts later; posting anywhere on an emitter (AudioEmitter.Post(evt, localOffset, values), AudioEvent.PostAtOffset, AudioManager.Post(evt, emitter, localOffset, values)) and moving a playing sound (AudioHandle.SetLocalOffset / SetWorldPosition); timed posts (Post(emitter, startDsp, startOffsetSeconds, values) on the event, emitter and manager) and AudioEvent.EstimateDuration (AudioDuration); Pepper’s Doppler (AudioContainer.Doppler, DopplerExaggeration, DopplerMaxCents, DopplerMode, DopplerMath), AudioEmitter.Teleport, AudioListeningPoint.ClearVelocity; ducks combining by the deepest (AudioBus.ApplyDucking); AudioState.VolumeMode (StateVolumeMode: SetLevel, Offset), AudioManager.ClearState, AudioBus.StateOffsetDb / PlayerVolumeDb / IsMuted, AudioManager.MasterVolumeDb / SetMasterVolumeDb / IsAllMuted; SetBusVolume with an unknown name changes nothing.
• Many cars and split screen (3.1.0-dev.8): AudioManager.SpatialMode (AudioSpatialMode: Unity, Pepper), Pepper panning and attenuating 3D sounds from every listening point by its Weight and PanWidth (GainStack.SpatialGain), reproducing Unity’s 3D at one point; AudioManager.SetPrimaryListeningPoint / PrimaryListeningPoint, AudioListeningPoint.IsPrimary; a point whose transform is destroyed is removed; each play measured from its own point, gliding over 100 ms when it changes; AudioBus.MaxRealVoices / RealVoiceCount; AudioManager.StealFadeSeconds (steals fade); Low priority weighs 1/8 in importance; every pooled voice carries its low- and high-pass from the start; AudioVoiceDebugInfo.listeningPoint / pepperPanned / spatialGainDb / pan.

v3.0.0 - July 2026

• New subsystem: Multiplayer (Netcode) — backend-agnostic networked audio. Post a sound on one machine and every player hears it; no audio crosses the wire, only a ~36-byte descriptor each machine plays locally. Install one netcode stack and its adapter lights up automatically; a backend you didn’t install never compiles. See Manual chapter 17 and API → Netcode (Multiplayer).
  • NetworkedAudioManager façade — PostNetworkedEvent / PostNetworkedLoop / StopNetworked, server-authored PostServer*, scheduled/quantized cues
  • Emitter binding (sound follows a moving NetworkObject), shared loop lifetime + late-join snapshot, cross-client variation parity (seed on the wire)
  • Competitive hardening — server authority/validation, proximity culling (= don’t-send), per-event Unreliable Delivery
  • Musical sync — NetworkClock (network-time ↔ local dsp) + NetworkMusicTransport (tempo/phase), loop phase-lock tiers (Tier 2 experimental — not for production yet)
  • NetworkAudioSettings scene component; Network Audio Debug window; per-event Multiplayer foldout on AudioEvent
  • Adapters: NGO 2.x (SFXSYSTEM_NGO) and FishNet v4 (FISHNET). Verified on FishNet, one PC and two PCs; NGO has had light testing only.
• New subsystem: Audio Loading & Memory — tiered, opt-in control over clip RAM footprint and load timing. Default direct-reference path is unchanged and zero-overhead. See Manual chapter 16 and API → Loading.
  • AudioClipWarmer + AudioManager.WarmClip/WarmClips/WarmContainer — front-load clip audio data ahead of first play (no extra package)
  • IAudioClipSource seam (AudioClipSources.Active, default DirectClipSource) — containers resolve clips through it; per-entry address on WeightedAudioClip / SequenceEntry / RoutingEntry
  • Optional Addressables backend (gated on com.unity.addressables): load/unload a region’s clips by zone via ZoneAudioBank + AddressableAudioRuntime, voice-safe refcounted residency (banks + voices + pins), drop-on-cold, linger
  • Addressable Audio Authoring window (convert clips in one click) + Loading Validator window. Voice-hold correctness verified; build-scale RAM profiling pending.
• Additive — no v2.5.0 APIs changed. Both subsystems are inert until opted into (netcode: install a stack + add a transport; loading: use the warmer or install Addressables).

v2.5.0 - May 2026

• New subsystem: Occlusion Mixer Slot Pool — per-bus pre-built mixer-slot pools replace per-source AudioLowPassFilters. Eliminates the per-acquire biquad-reset click; cutoff modulation is a stable mixer-side write. See Manual chapter 14.
  • OcclusionLayout + OcclusionSlot types
  • AudioManager.AcquireOcclusionSlot / ReleaseOcclusionSlot and slot-pool diagnostics
  • Multi-ray partial occlusion (occlusionRayCount, occlusionRaySpreadMeters) — N-level edge gradient instead of binary
  • Per-frame exponential smoothing via occlusionGainSmoothingSeconds / occlusionCutoffSmoothingSeconds
  • Occlusion Layout Builder window (Window > Audio System > Occlusion Layout)
• New subsystem: Reverb Send Buses & Per-Zone Reverb — source-side routing + listener-side bus-character driver. SFX voices contribute to the reverb of the room they’re fired in; each bus’s character is defined by its ReverbSendBus asset (the single source of truth, since the 2026-05-21 unification). See Manual chapter 15.
  • ReverbSendBus (one acoustic-space identity per asset) + ReverbSendBusRegistry
  • AudioReverbProfile (six SFX Reverb character params, authored on the bus SO)
  • AudioZone.ReverbBus (routing target; character lives on the ReverbSendBus SO, not the zone)
  • Convolution reverb option (Phase 3.2, Windows x64, beta) — ReverbSendBus.UseConvolution swaps the parametric SFX Reverb for a native impulse-response convolution plugin. See Manual §15.14.
  • AudioManager.WriteReverbSendsImmediate — synchronous on-Play write so first audio frame carries correct routing
  • AudioContainer opt-in flags: AllowReverbSend, ReverbSendLevelDb, StaticEmitter, ExplicitReverbBus
  • Reverb Send Buses window (Window > Audio System > Reverb Send Buses) + per-asset Generate/Validate/Apply/Pull inspector
• AudioContainer routing flags — UseOcclusion, UsePropagation, AllowReverbSend opt SFX containers into the new slot pool. UseOcclusion enables raycast occlusion; UsePropagation composes propagation cutoff into the same voice via most-muffled-wins (min cutoff, multiplicative gain). The two routing/muffling subsystems now integrate cleanly with the propagation subsystem from v2.3.0.
• Soft zone boundaries — AudioZone.entryFadeMeters smooths reverb sends, listener-side reverb character, and BaseVolumeDb across the zone surface. 0 (default) preserves the v2.3.0 binary boundary.
• Outside-the-graph behavior — PropagationManager.silenceOutsideGraph toggle picks between whitelist mode (outside = audible, default) and complete-coverage mode (outside = silent + muffled).
• Reachability-preserving culler — PropagationProximityCuller now keeps zones reachable through near portals active even when individually out of activation-radius range. Two-zone-via-portal scenes no longer fall silent on tight radii.
• Higher default solve rate — PropagationManager.solveRateHz default 8 → 15 (≈ 67 ms cache step). Fast zone-trigger crossings and partially-closed-portal transitions no longer surface audible step changes.
• Lower default open low-pass — AudioPortal.openLowPassHz default 18000 → 12000. A portal without a door source is now audibly distinct from full transparency on typical speakers, surfacing authoring mistakes early.

v2.3.0 - April 2026

• New subsystem: AudioSystem.Propagation — zone/portal graph-based routing for ambient beds
• AudioZone / AudioPortal MonoBehaviours (authoring)
• AmbientSource / AmbientEmitter (declaration + pooled persistent voice)
• PropagationManager singleton + AudioListenerZoneTracker
• IPortalDoorSource interface for portable door integration
• PropagationProximityCuller for large-level scale (per-node activation radius + global caps)
• Pure Dijkstra solver with reusable buffers (allocation-free hot path)
• Blend-region crossfading for pop-free doorway traversal
• Additive — does not modify any v2.2.0 APIs

v2.2.0 - March 2026

• Runtime container override system (3D settings)
• Per-voice control on AudioMultiHandle
• RTPC transitions with TransitionRTPC()
• StopAllSounds(), PauseAll(), UnpauseAll()
• MuteAll(bool) replaces separate Mute/Unmute methods
• AudioStatistics replaces Statistics struct
• VoiceStealBehavior: added Furthest, removed None
• GainStack: added DuckingGain, ApplyToSource(), Reset()
• SequenceContainer: AutoAdvance, PlayNext()
• AudioBus: ducking system, effect sends
• AudioState: structured property classes
• Multi-position emitter components
• IRTPCListener interface
• ListenerUtil utility class
• AudioVoiceDebugInfo for runtime debugging

v2.0.0 - January 2026

• AudioEventRegistry system
• Quick Sound Setup Wizard (5 modes), now Sound Builder (Simple)
• AudioEvent Inspector validation
• Batch import support

v1.0 - Initial release

• Complete audio middleware system
• 5 container types
• Event system with multi-actions
• Hierarchical bus mixing
• State management
• Multi-position audio
• Voice virtualization & LOD
• Occlusion system

---

End of API Reference

---

For tutorials, see Cookbook in the User Guide For deep knowledge, see Manual in the User Guide For quick start, see Quick Start in the User Guide

Pepper.SFX v3.0.0 - Professional Audio Middleware for Unity