Game Development

Audio in Games

Research shows: turning off sound drops emotional engagement in games by 30%. Half-Life 2, The Last of Us, Red Dead Redemption 2 - games that shaped the emotional memory of a generation, in large part because of sound.

**Red Dead Redemption 2** (Rockwell): 60,000+ audio assets, Wwise, procedural ambient sound system for each ecosystem - every region sounds unique
**Microsoft Flight Simulator 2020**: procedural engine sound generation through physical models - not recorded samples, but synthesis based on thrust, RPM and air pressure data
**Gran Turismo 7**: 400+ cars, each with a unique sound across all operating modes - impossible without procedural synthesis and FMOD automation

Spatial Audio: HRTF and 3D Positioning

2004. Half-Life 2. Valve designer Marc Laidlaw wrote in a post-mortem: 'Sound turned the shooter into a tactical one'. Gunfire behind a corner is audible. Grenades are heard before they are seen. Spatial audio turned the game into an information system.

HRTF (Head-Related Transfer Function) is a mathematical model of how the ear perceives sound from different directions - filtered through the shape of the pinna, head, and torso. The result: the brain identifies the 3D position of a source through headphones. Microsoft Spatial Sound, Dolby Atmos for gaming - all based on HRTF.

ML application: Valve Research (2023) demonstrated a system for generating personalized HRTF from a photo of the user's head. Personalized HRTF improves sound localization by 40% compared to an averaged model. Architecture: photo encoder + MLP predicting filter coefficients.

What does spatialBlend = 1.0f mean for a Unity AudioSource?

Adaptive Music: States and Transitions

The Last of Us Part II. When Joel and Ellie hide from infected, the music changes. Not just a fade - it restructures in real time: drums disappear, leaving only a low-frequency bed. Then conflict - and immediately the full orchestra. This is not editing tracks. This is adaptive music.

Adaptive music runs on a finite state machine. Combat, stealth, exploration, cutscene - each state has its own music layer or theme. Transitions between them are synchronized to musical beats (bars, beats) to avoid jarring cuts. The Wwise Music Switcher implements this through hierarchical Music Segment groups.

Why are adaptive music transitions synchronized to musical beats rather than happening instantly?

SFX: Procedural and Sampled Sounds

Sound effects in games fall into two classes. Sampled: recorded real-world sounds (footsteps, gunshots, voices). Procedural: synthesized in real time (car engine, wind, water). Gran Turismo 7 synthesizes the sound of each of its 400+ cars procedurally - storing 8000+ hours of audio for all speeds and loads is impossible otherwise.

Randomization is the key to living SFX. The same footstep sound 50 times in a row is instantly recognizable monotony. The solution: pitch variation (+-5-10%), volume variation (+-3dB), random selection from a pool of variants (4-8 recordings of the same sound). Call of Duty has 6-8 footstep variants for each surface material.

Why is pitch variation (random changes in sound pitch) used in game SFX?

FMOD and Wwise: Middleware for Serious Projects

Red Dead Redemption 2. 300+ sound designers. 60,000+ unique audio assets. 5 years of development. Rockstar uses Audiokinetic Wwise. Not because built-in tools do not work - but because at that scale professional tools with visual authoring, hot-reload and a profiler are necessary.

FMOD Studio and Wwise are the two market leaders for game audio middleware. Both provide: a visual editor for audio events and states, RTPC (real-time parameter control), dynamic mixing, spatial audio processing, and a profiler. FMOD: easier to learn, better documentation, slightly cheaper. Wwise: more powerful for large AAA projects, interactive music, HDR audio.

Pricing: FMOD Studio is free up to USD 200K revenue, then licensed. Wwise is free for projects with up to 200 sounds, then licensed. Most AAA titles (BioShock to Cyberpunk 2077) use one of them. Built-in engine audio (Unity Audio, Unreal Audio) is the right choice for indie and smaller projects.

FMOD and Wwise are mandatory for any professional game

For indie games and projects with a small audio budget, built-in Unity/Unreal solutions are fully sufficient. FMOD/Wwise are justified with a large sound design team and complex interactive systems

FMOD/Wwise add integration complexity and licensing cost. Celeste, Hollow Knight, Stardew Valley achieved excellent audio without them. Hades uses FMOD. The choice depends on team size and audio complexity

What is the main advantage of FMOD/Wwise over built-in Unity/Unreal audio for a large project?

Key Ideas

**Spatial Audio**: HRTF + distance attenuation + Doppler. spatialBlend=1.0 in Unity. Personalized HRTF improves localization by 40%.
**Adaptive Music**: finite state machine + Music Switcher. Beat-synced transitions. RTPC for dynamic parameters (intensity, enemy proximity).
**SFX design**: sampled vs procedural. Pitch/volume randomization + variant pool. Procedural engine sound via RPM and throttle parameters.
**Middleware**: FMOD Studio (simpler) vs Wwise (more powerful for AAA). Key advantage: visual editor with hot-reload for sound designers.

Вопросы для размышления

How does turning off sound affect the experience of horror games? Why is sound design more critical for horror than for action?
Procedural engine sound vs recorded samples: where is the realism boundary? What parameters cannot be conveyed procedurally?
FMOD/Wwise add a third-party middleware dependency. How does this affect long-term game support during engine updates?

Связанные уроки

gd-14 — Game engine basics are needed to understand the audio pipeline
gd-16 — Audio UX is part of the overall Game UI/UX system
gd-12 — Physics in games affects audio - reverb, materials, collisions
net-03-physical
la-01-vectors-intro