A horror game is less a visual experience than an acoustic trap. The player looks at the screen, but the body responds to whatever signal arrives at the ears. The image presents the frame you expect; the audio implies the frame you don't. Sound design is therefore not decoration in horror, it is a mechanical backbone. A wrongly mixed audio bed can turn the most beautiful scene into an over-bright commercial break.
At Althera Games, we build NightRecord: Thin Walls's Soviet apartment atmosphere on top of audio decisions, before any visual decision. In this article we open up indie horror sound design end to end, from binaural fundamentals to the FMOD versus Wwise call, from the architecture of silence to concrete examples we use in NightRecord. For the atmosphere side of horror, our psychological horror design article and environmental storytelling guide are useful companion reads.
Why Sound Is Fear
The human visual system has to turn its eyes toward something to function. The ear is open in every direction at once. When you hear a tap behind you in a dark room, your brain initiates a freeze reflex before it has even rotated your head toward the source. That reflex is the raw material of horror design: an uncertainly localized sound does in half a second what visual storytelling can't manage in minutes.
The neurological basis sits with the amygdala and its short auditory pathway. Visual data takes a longer trip through the cerebral cortex; certain auditory signals drop straight into the amygdala. Your body has already responded before you have consciously decided what is "scary." Horror designers target that short pathway. Frequency choice, dynamic range, and spatial localization are tools aimed at the body, not the conscious mind.
A practical observation: the most common moment players abandon a horror game is rarely something on screen. It's usually a sound they could not predict, fired one too many times. Your job as designer is to manage that fatigue. Sustained high tension is unsustainable; you need to set a breathing rhythm between silences and bursts. The tempo of that breath is set by audio.
Binaural Audio and HRTF
Stereo is the crudest form of sending two different signals to two ears. You pan a sound left, and the player hears it on the left. In the real world, however, a sound source is not just left or right; it is above, below, in front, behind, or behind a half-closed door. That positional information is carried by the anatomy of your head: HRTF (Head-Related Transfer Function) is the filter that describes how sound transforms as it passes around your pinnae, skull, and shoulders.
Binaural audio applies that HRTF transformation digitally inside the engine. The result lets a headphone listener feel the direction of a sound bodily. In a NightRecord corridor scene, when a footstep crosses through the apartment behind the player, the player has to physically turn their head. That reflex is unobtainable from simple stereo panning; it comes from HRTF. Steam Audio and Resonance Audio, both free, are two strong HRTF plugins that work inside FMOD and Wwise.
Practical advice: enable HRTF only on 3D sounds close to the player character. Music and broad ambient layers should stay stereo, since those layers already produce a non-positional feel. Otherwise the HRTF processor burns CPU unnecessarily and music acquires an unstable spatial impression. The relationship between your character's listener component and your sound sources has to be set up correctly in the engine, or binaural's added value disappears. For a parallel argument about "physical credibility," our Lumen guide applies the same logic to lighting.
FMOD vs Wwise Decision Matrix
The audio middleware choice is one your game lives with for life; you can change it later, but the cost is real. The two big options are FMOD and Wwise. Both ship official integrations for Unreal Engine 5, Unity, and many other engines.
FMOD has a leaner toolset. Its authoring app, FMOD Studio, is intuitive for composers and sound designers; parameters, snapshots, and event handling are presented clearly. Its licensing is unusually friendly to indie projects: free for projects with under $200,000 USD in gross revenue per calendar year, and tiered above that with reasonable rates for indie budgets.
Wwise is a much wider tool ecosystem. Features like RTPCs (Real-Time Parameter Controls), States, Switches, advanced voice management, motion plug-ins, and dynamic mixing are industry standard for larger productions. That richness has a price: the learning curve is steep, and small teams have hundreds of features they will never use. Licensing on indie tiers is more limited, but the Indie Profile program offers a free entry point.
A practical decision matrix: pick whichever your team has already used. If nobody has used either, FMOD's learning ramp is one to two weeks shorter and creates no licensing pressure on indie budgets. If your project carries large-scale mixing requirements, or you intend to hire an audio programmer down the line, Wwise's investment value is high. Whichever you choose, have someone on the team prototype the middleware end to end on a tiny test project before committing; that prototype will clarify integration quality with your engine.
The Architecture of Silence
The least understood facet of horror is the design of silence, not loudness. Tiring a player with screams is easy; forcing them to hear their own breath is the actual craft. Silence is not a mistake; silence is a tool.
In practice, three kinds of silence exist. Spatial silence: the room's natural acoustic floor. Story silence: a silence that signals a character or event is absent. Wrong silence: the deliberate withdrawal of sounds the player expects. The third is the strongest weapon in horror. Walking through a corridor, if the distant city hum you've been hearing suddenly cuts out, no jump-scare matches that effect. That is one of the techniques we use throughout NightRecord's Soviet apartment.
Building silence starts with carefully labeling your audio layers. Every scene runs at least four layers: an ambient bed (the room's underlying hum), spot ambient (a distant car, a door), character foley (footsteps, breath, clothing), and a threat layer (enemy or tension cues). Silence design is the deliberate withdrawal of one of those layers, or all of them at once. In FMOD you manage that with snapshots, in Wwise with states.
Silence is not a sound you can't add; it is a sound you choose to remove. In horror, tension comes not from what you add but from what you take away.
Enemy Sounds and Spatial Cues
An enemy in a horror game is not just an asset, it is an acoustic personality. The player should hear the creature before seeing it; from the sound, they should quickly understand who it is, where it's coming from, and how far away it is. That is a complex balancing problem for the sound designer.
When layering creature audio, three components are critical. Action sound: what it tells you the creature is doing (footsteps, breath, scraping). Character sound: who it is (growl, scream, murmur). Atmosphere sound: how close it is (low-frequency rumble, air vibration). When all three layers run together, three separate systems in the player's brain engage: localization, recognition, tension. If a layer is missing under the enemy, the horror feels incomplete.
For localization, attenuation curves are the central tool. If a creature is audible from 30 meters, the player tracks it for far too long and the psychological pressure dissipates. If it's audible at 5 meters, surprise compounds. In NightRecord, when a breath sound comes from the end of a corridor, the attenuation curve starts at 4 meters and is fully cut by 8 meters. Adding a distance-based low-pass filter further pushes physical credibility: distant sounds naturally lose their high frequencies, near sounds sharpen.
Music vs Ambient
The most common debate in horror is whether to use music at all, or to lean entirely on ambient. The answer is not categorical. Music is a commentary tool: it tells the player "you are tense right now." Ambient is a presence tool: it tells the player "you are here right now." The ratio between them defines the genre of your horror.
High-music horrors (the Resident Evil family) are paced and cinematic; they put the player on an emotional train. Low-music horrors (Amnesia, SOMA) build a settled sense of reality; when the player walks into the next room, they feel they are still in the same world. NightRecord sits in the second camp. We bring music in only at specific narrative anchor points; the rest of the runtime flows as a detailed ambient texture.
The technical translation is to set up FMOD's tempo aware music system together with an ambient stem mix. When music plays, the ambient mix drops by -3 dB; when music ends, ambient crossfades back up smoothly. This simple piece of automation lets the player subconsciously distinguish narrative moments from atmosphere moments without you ever signaling it explicitly.
Practical Examples From NightRecord
NightRecord: Thin Walls's central setting is the small Soviet apartment Vadim leaves to his wife. Built in the Khrushchyovka style, this building's acoustics are the foundation of the game's atmosphere. The sound transmission through concrete walls, the creak of old parquet, the metallic complaint of the heating pipes; each of these was treated as its own character.
The apartment's sound map has four primary zones. Inside the flat: soft, enclosed, intimate acoustic. Corridor: reverberant, mid-cool frequency balance, concrete-led. Stairwell: the most reverberant zone, low frequencies amplified, distant footsteps boosted. Street: wind, distant car, a slamming door. The transitions between these zones are not smooth; when a door opens, the acoustic contrast directly anchors the player into the new space.
Through FMOD snapshots, we run a mix that drifts with the time of day. At 7:00 in the morning, the apartment's "life mix" is on: a child's voice from below, a radio next door, footsteps overhead. By 22:00 those layers have slowly retracted. At 03:00 only the heating pipe's complaint, a distant street dog, and the character's breath remain. In that "wrong silence" state, a normally invisible door tap can stop the player's breath. Sound design is the story.
For enemies, NightRecord doesn't field a classic creature; absence itself is the threat. That is an interesting problem for sound design: how do you craft the audio of something you cannot see and cannot name? Our solution was to treat the building itself as the creature. The pipes, the elevator mechanism, the kettle's whistle, the cabinets that move on their own; all of them are processed at low levels and on late-onset attenuation curves. The player turns toward the tap, walks toward it, and finds nothing. Horror arises from a non-visual causality.
For mixing on a small team, our advice: build a reference mix in Reaper or Cubase first, and have all sound designers work down the same master path. Aim for an average master meter of around -23 LUFS; keep dialogue at -18, music at -16, ambient at -28. Without that numerical discipline, horror's loud moments become destructive and quiet moments become inaudible. Also, every weekend, run a short play session on a different sound card and a different pair of headphones; an entire team listening through the same chain is the single biggest threat to consistent horror. Our UE5 indie development article adds perspective on team workflows; a few examples from NightRecord's audio lab live on our games page.
Frequently Asked Questions
Should I choose FMOD or Wwise for indie horror?
For small teams, FMOD is usually the more practical pick: a gentler learning curve and indie-friendly licensing, including a free tier for projects under $200,000 USD in gross revenue. Wwise has a richer toolset and is the industry standard for mid-to-large productions, but it has a steeper ramp. Before deciding, evaluate the integration quality with your engine, your team's prior experience, and the actual feature set you will end up using.
Why is binaural audio so important in horror?
Binaural audio respects the fact that the human ear is a localization system built from two microphones, and it places sound in real 3D space around the listener's head. Horror is a response to unknown direction; when a player can pinpoint exactly where a sound is coming from, tension drops, but when they can't, the body shifts into a defensive posture. HRTF-based binaural processing presents a footstep behind the room with centimeter-level believability for headphone players, and that physical credibility is the foundation of horror sound design.
How do you test sound mapping and mix on a small team?
Step one is to play in at least three listening environments: studio headphones, consumer headphones, and laptop speakers. Step two is measuring loudness numerically; the LUFS metric is especially useful, and a target of around -23 LUFS keeps horror's loud peaks under designer control. Step three is player testing: measure tension behaviorally, not verbally. When does the player pause, retreat, or pull off their headphones? This three-layered test lets small teams ship a stable mix without a full-time mix engineer.
How should I work with a freelance composer?
Start your search in the right places: Soundcloud, Bandcamp, and the Discord communities for indie game composers all surface candidates whose catalogue you can listen to. In the outreach, share aesthetic references for the project rather than a generic application; tonal fit matters as much as skill. In the contract, clarify the rights model (work-for-hire, royalty, or hybrid), the deliverable formats (stems, master, loop points), and the revision policy in writing. Start the budget with a clear range; an open-ended brief often produces mutual disappointment.
How do you plan an audio mix budget on an indie project?
A pro mix engineer's daily rate can be high, but a practical compromise on indie projects is to split mixing into two phases: an in-team 'playable mix' you maintain throughout production, and a short professional pass before launch. Booking that second phase as a 3-5 day consult lifts the audio's mastering quality dramatically even on a small budget. Learning FMOD and Wwise's snapshot systems lets you make half of the mix decisions in-game before the engineer ever touches the project.
Conclusion: Designing Silence
In horror, sound is not a coating, it is a skeleton. It carries half the atmosphere, signals the entire story, and builds a cognitive bridge between the player's body and the game. Binaural systems handle physical credibility; FMOD and Wwise bind that credibility to game state; the architecture of silence governs the rhythm of horror. All three layers are within reach of indie teams, free or under reasonable licenses.
In NightRecord: Thin Walls, the apartment's "wrong silence" became a manifesto for our sound design. The strange stillness that descends on the building after Vadim leaves is something no visual tool can describe. If you are working on a horror project, design what the player will not hear in the next scene, more than what they did in the previous one. Our environmental storytelling article is the visual cousin of this grammar; we recommend reading the two together.