## Phase 167: Neith Engine Audio Architecture Depth — Wwise / FMOD / Unreal MetaSounds / Unity Audio + Vivox / Dolby Atmos Runtime / DTS:X / Steam Audio / Oculus Audio SDK / Resonance / Project Acoustics Parity

Closes the engine-audio sovereignty gap. Phase 132 ships audio drivers and
plugin hosting. Phase 133 ships the Composer DAW. Phase 38 ships Euterpe music
intelligence. None deliver a **game-engine audio runtime** — the bus graph,
spatialisation, HRTF, reverb, procedural music, dialog system, and cinematic
mixer that sits between the sound designer and the end player. Godot audio has
~20 tasks and many features are light across phases; this phase makes engine
audio first-class.

Sovereign replacement for Audiokinetic Wwise (banks, RTPCs, states, switches,
events, SoundSeed, Motion, Spatial Audio, Strata, Impacter), Firelight FMOD
Studio + Core, Unreal MetaSounds + Quartz + Submixes + SoundClasses, Unity
Audio + Audio Random Container + Vivox, Dolby Atmos game runtime, DTS:X, Valve
Steam Audio, Meta Oculus Audio SDK, Google Resonance Audio, Microsoft Project
Acoustics, Convolvr, Hexany Audio / Elias adaptive music.

---

### 167.1 @neith/audio-graph — Node Graph & Runtime

- [x] **167.1.1** Directed audio graph (source → buses → output)
- [x] **167.1.2** Dynamic graph mutation at runtime
- [x] **167.1.3** Per-bus sample-rate / channel-count conversion
- [x] **167.1.4** Sample-accurate scheduling
- [x] **167.1.5** Sub-frame tick support (per-buffer)
- [x] **167.1.6** Cross-fade / gain automation
- [x] **167.1.7** Snapshot / state presets
- [x] **167.1.8** Multi-threaded decode
- [x] **167.1.9** Audio thread priority / real-time hints
- [x] **167.1.10** Glitch-free add / remove node
- [x] **167.1.11** Bus sidechain routing
- [x] **167.1.12** Auxiliary send / return buses
- [x] **167.1.13** Solo / mute bus with cascade
- [x] **167.1.14** Hierarchical groups (dialogue / music / SFX / UI)
- [x] **167.1.15** Hot-reload of graph asset

---

### 167.2 @neith/audio-source — Source Playback Types

- [x] **167.2.1** Sample playback (WAV / FLAC / Opus / Vorbis)
- [x] **167.2.2** Streamed long-form audio
- [x] **167.2.3** Procedural oscillator / noise generator
- [x] **167.2.4** Physical modelling source (string / drum / wind)
- [x] **167.2.5** Granular synthesis
- [x] **167.2.6** Concatenative (Phase 133 reuse)
- [x] **167.2.7** MIDI instrument (sampler / synth)
- [x] **167.2.8** Ambisonic source
- [x] **167.2.9** Impulse-response convolution source
- [x] **167.2.10** Random container (weighted / sequential / shuffle)
- [x] **167.2.11** Switch container (context-driven)
- [x] **167.2.12** Sequence / layer container
- [x] **167.2.13** Stinger / one-shot
- [x] **167.2.14** Crossfade playlist
- [x] **167.2.15** Time-stretch + pitch-shift per source

---

### 167.3 @neith/audio-spatial — 3D Spatialisation

- [x] **167.3.1** 3D positional attenuation curves
- [x] **167.3.2** Distance attenuation with inverse / linear / custom
- [x] **167.3.3** Directional emitter cone
- [x] **167.3.4** Doppler with vmax clamp
- [x] **167.3.5** HRTF binaural (multi-HRIR database)
- [x] **167.3.6** Personalised HRTF import
- [x] **167.3.7** Near-field HRTF extension
- [x] **167.3.8** Ambisonic (1st / 3rd / 5th order) decode
- [x] **167.3.9** Atmos bed + object routing
- [x] **167.3.10** DTS:X object metadata
- [x] **167.3.11** Speaker panning 5.1 / 7.1 / 7.1.4 / 9.1.6
- [x] **167.3.12** Vector-base amplitude panning
- [x] **167.3.13** Headphone / speaker auto-switch
- [x] **167.3.14** Head-locked / world-locked mode
- [x] **167.3.15** Ear-to-ear ITD / ILD modelling

---

### 167.4 @neith/audio-environment — Acoustics & Environmental FX

- [x] **167.4.1** Environment / reverb zones
- [x] **167.4.2** Convolution reverb with IR library
- [x] **167.4.3** Algorithmic (FDN / Schroeder) reverb
- [x] **167.4.4** Occlusion + obstruction ray cast
- [x] **167.4.5** Transmission through materials (wood / glass / concrete)
- [x] **167.4.6** Diffraction around corners
- [x] **167.4.7** Pre-baked wave-acoustics (Project Acoustics-class)
- [x] **167.4.8** Real-time ray-traced acoustics (Steam Audio-class)
- [x] **167.4.9** Reverb morph between zones
- [x] **167.4.10** Weather / wind ambience
- [x] **167.4.11** Crowd / roomtone layering
- [x] **167.4.12** Underwater filter
- [x] **167.4.13** Tunnel / canyon early reflection
- [x] **167.4.14** Hearing-damage / concussion filter
- [x] **167.4.15** Vehicle interior acoustics

---

### 167.5 @neith/audio-mix — Mixer, Ducking & Snapshots

- [x] **167.5.1** Real-time mix snapshot interpolation
- [x] **167.5.2** Sidechain ducking (music under dialogue)
- [x] **167.5.3** Dynamic mix profiles (combat / dialogue / cinematic)
- [x] **167.5.4** Priority-based voice culling
- [x] **167.5.5** Bus DSP chain (EQ / compressor / limiter / saturator)
- [x] **167.5.6** True-peak limiter with ISP oversample
- [x] **167.5.7** LUFS live loudness metering
- [x] **167.5.8** Loudness normalisation (ITU-R BS.1770)
- [x] **167.5.9** Dialogue intelligibility booster
- [x] **167.5.10** Mix sheet export for audio lead
- [x] **167.5.11** Scene-specific sub-mix
- [x] **167.5.12** Multi-listener (split-screen) mix
- [x] **167.5.13** Cinematic vs gameplay auto-bias
- [x] **167.5.14** Accessibility: mono down-mix + clarity
- [x] **167.5.15** Visual mixer UI with metering

---

### 167.6 @neith/audio-dialogue — Dialogue System

- [x] **167.6.1** Dialogue line database
- [x] **167.6.2** Character voice bank (Phase 162)
- [x] **167.6.3** Localisation per language + region
- [x] **167.6.4** Conditional line selection rules
- [x] **167.6.5** Barks / triggered short lines
- [x] **167.6.6** Conversation flow manager
- [x] **167.6.7** Branching dialogue system
- [x] **167.6.8** Radio-comm / walkie voice FX
- [x] **167.6.9** Procedural voice (TTS) fallback (Phase 162)
- [x] **167.6.10** Lip-sync phoneme export
- [x] **167.6.11** Subtitle / caption trigger (a11y)
- [x] **167.6.12** Speech-to-visemes at runtime
- [x] **167.6.13** Emotion tag drives mix
- [x] **167.6.14** ADR / dubbing pipeline integration
- [x] **167.6.15** Scripted-vs-emergent chatter

---

### 167.7 @neith/audio-music — Adaptive Music System

- [x] **167.7.1** Vertical / horizontal layers
- [x] **167.7.2** Musical bar-/beat-aware transitions
- [x] **167.7.3** Stinger triggering on cue
- [x] **167.7.4** Tempo / key-aware crossfade
- [x] **167.7.5** Generative stems from Phase 38 Euterpe
- [x] **167.7.6** Mood graph (exploration → combat → epilogue)
- [x] **167.7.7** Tension parameter coupling
- [x] **167.7.8** Real-time mixing of live-composed music
- [x] **167.7.9** Theme motif retrieval system
- [x] **167.7.10** Ableton Link clock sync (Phase 133)
- [x] **167.7.11** MIDI-driven in-game instrument
- [x] **167.7.12** Procedural percussion for footstep
- [x] **167.7.13** Muffled / distant music FX
- [x] **167.7.14** Audio-reactive VFX binding (Phase 161)
- [x] **167.7.15** Licensing-aware cue-sheet export

---

### 167.8 @neith/audio-synth — Procedural Audio / SFX

- [x] **167.8.1** Impact / breakage synthesis (modal)
- [x] **167.8.2** Footstep procedural with material / weight
- [x] **167.8.3** Weapon / gun SFX generator
- [x] **167.8.4** Vehicle engine (granular + additive)
- [x] **167.8.5** Wind / whooshes
- [x] **167.8.6** Water splash / flow
- [x] **167.8.7** Fire / electrical buzz / thunder
- [x] **167.8.8** Creature vocal synth
- [x] **167.8.9** UI sonification toolkit
- [x] **167.8.10** Foley assistive (from animation curves)

---

### 167.9 @neith/audio-vocie — Voice Chat & Mic Integration

- [x] **167.9.1** Opus voice chat codec (reuse Phase 159)
- [x] **167.9.2** Spatial proximity voice
- [x] **167.9.3** Team channels
- [x] **167.9.4** Noise suppression (RNNoise-class)
- [x] **167.9.5** Echo cancellation
- [x] **167.9.6** Gain-normalise
- [x] **167.9.7** Voice changer / masking
- [x] **167.9.8** Speech-to-text moderation
- [x] **167.9.9** Mute / block / transcript
- [x] **167.9.10** Push-to-talk vs voice activated

---

### 167.10 @neith/audio-profiler — Audio Profiler & Diagnostics

- [x] **167.10.1** Voice count + slot usage (Phase 160 integration)
- [x] **167.10.2** DSP budget per bus
- [x] **167.10.3** Plugin-host CPU cost
- [x] **167.10.4** Buffer-underrun events
- [x] **167.10.5** Mix-graph flamegraph
- [x] **167.10.6** Asset-load hitches
- [x] **167.10.7** Dialogue / music / SFX stem usage
- [x] **167.10.8** Latency trigger-to-sound meter
- [x] **167.10.9** Waveform scope view
- [x] **167.10.10** Reverb / occlusion cost breakdown

---

### 167.11 @neith/audio-tooling — Audio Designer Workflow

- [x] **167.11.1** Event Designer (Wwise-class authoring UI)
- [x] **167.11.2** State / switch / RTPC editor
- [x] **167.11.3** Timeline action graph (MetaSounds-class)
- [x] **167.11.4** Bank / soundbank packaging
- [x] **167.11.5** Asset browser with waveform
- [x] **167.11.6** Offline bake of conditional content
- [x] **167.11.7** Localisation matrix view
- [x] **167.11.8** Compare A/B of two mixes
- [x] **167.11.9** Playthrough recorder for review
- [x] **167.11.10** Live-connect to running game
- [x] **167.11.11** Edit while game runs (tweak live)
- [x] **167.11.12** Version history per asset
- [x] **167.11.13** Snapshot & rollback of scene mix
- [x] **167.11.14** Import Wwise project
- [x] **167.11.15** Import FMOD project

---

### 167.12 @neith/audio-accessibility — Accessibility

- [x] **167.12.1** Mono down-mix toggle
- [x] **167.12.2** Dialogue clarity boost
- [x] **167.12.3** Haptic substitute for sound
- [x] **167.12.4** Captions (full subtitle library)
- [x] **167.12.5** Directional-sound visual indicator
- [x] **167.12.6** Colour-coded sound source HUD
- [x] **167.12.7** Left-right balance slider
- [x] **167.12.8** Hearing-aid mode (speech EQ boost)
- [x] **167.12.9** Volume ceiling protection
- [x] **167.12.10** Dyslexia / cognitive fonts for captions

---

### 167.13 Testing & Validation

- [x] **167.13.1** Voice-count stress (≥ 512 concurrent voices)
- [x] **167.13.2** Latency < 20 ms P99 trigger-to-ear
- [ ] **167.13.3** HRTF ABX vs a reference renderer — the fail-loud evidence
      grading is complete. SteamAudio and OpenAL-Soft are open-source HRTF
      references that install freely, so the objective comparison is actionable;
      but the ABX itself is a double-blind human listening panel on calibrated
      headphones, which only people can perform (stated 2026-09-18).
      `blocked:human`
- [x] **167.13.4** Atmos object renderer bit-compare
- [x] **167.13.5** Wwise / FMOD project import fidelity
- [ ] **167.13.6** 24-hour soak without underrun — the evidence grading enforces
      a live device, ≥ 86,400 s and zero underruns. Run it on the fleet's real
      audio device (CoreAudio on the Mac or ALSA on the Linux server). Verify: a
      24-hour continuous-render soak on a live audio device completes with zero
      underruns/dropouts and the grader accepts the captured duration and frame
      coverage.
- [x] **167.13.7** Multi-listener mix correctness
- [x] **167.13.8** Adaptive music transition test corpus
- [x] **167.13.9** Loudness compliance for TV broadcast
- [x] **167.13.10** Dogfood: Phase 42 engine demos + Phase 27 franchise
