Why your game feel is 80% sound

Players will forgive rough edges in a lot of places before they forgive a hit that doesn't sound like a hit. Audio is the fastest feedback loop a game has — faster than animation, faster than particles, faster than the UI catching up to what just happened.

“

On every project we start audio pass planning at the same time as combat prototyping, not after it. A placeholder thwack that lands on-beat teaches more about game feel than a finished animation with silence behind it.

The 100ms rule

Anything a player does — a jump, a parry, a menu tap — needs an audible response inside roughly 100 milliseconds or it reads as unresponsive, even if the underlying logic fired instantly. Visuals alone rarely close that gap fast enough on their own; a one-frame VFX pop can still feel mute without a transient sitting right on top of it. We treat that window as a hard budget, the same way we'd budget frame time, and test it with placeholder stub sounds before any art exists.

Layers, not loudness

The trick isn't more sound — it's layered sound with headroom. Stack a transient, a body and a tail, and let the mix breathe so the next hit still reads clearly instead of turning into noise. A typical impact in our library is three to four layers: a short high-frequency snap for the attack, a mid-range body that carries the weight, a low-end thump players feel more than hear, and a short tail that gives the ear somewhere to land before the next event.

Cranking every layer to full volume is the most common mistake we see in early builds. Once four or five systems are all fighting for the same headroom, everything blurs into a wall of noise and the moments that should feel big — a boss slam, a critical hit — stop standing out. Ducking and priority rules matter more than raw volume.

Prototype the sound before the animation

Most teams treat audio as a coat of paint applied once gameplay is "done." We flip that order wherever we can. A rough placeholder hit sound, even a mouth-clicked stand-in recorded on a phone, tells a designer more about whether an attack feels right than a week of animation polish with the sound channel muted. It's cheap to iterate on, and it exposes timing problems — a windup that reads as too slow, a recovery that feels unforgiving — long before an animator has sunk real hours into it.

Where the mix actually happens

A lot of the "sound design" that makes a hit feel good isn't the source recording at all — it's mix decisions made in real time as the game plays. Sidechain ducking pulls music and ambience down a few dB the instant a big hit lands, so the transient cuts through without anyone touching the fader. Priority systems decide which of twelve simultaneous gunshots actually gets voiced when the engine can only afford eight active channels. None of this is glamorous, but it's the difference between a mix that holds together under real combat load and one that turns to mush the moment four players fight on screen at once.

What we ship with

Our audio pipeline runs on FMOD for runtime mixing, parameter-driven layering, and the ducking/priority logic above — wired directly into whichever engine the project uses (mostly Unity or Unreal on our slate). Keeping the sound design in FMOD's event system rather than hard-coding playback in engine script means the audio team can rebalance a fight's entire mix without touching a line of gameplay code, and iterate at the same speed the design team does.