Your DAW Writes the Music. The Audio Engine Plays It.

You wrote the track in your DAW. It sounds finished. You drop the wav into the game, and it is wrong: it loops with a click, it fights the gunfire, it keeps playing its big emotional swell three seconds after the player already died.

Nothing is broken. You just handed a finished performance to a system whose entire job is to decide what to perform. A DAW and an audio engine are both "audio software" in the way a camera and a projector are both "photography equipment." Here is the actual difference, and what it changes about how you write.

A DAW is an offline tool with a fixed timeline

Ableton, FL Studio, Reaper, Logic, Cubase. Whatever you use, the deal is the same.

Time in a DAW is a line you already drew. Bar 33 happens after bar 32, every single time. Nothing in the world can interrupt it, so you can commit to everything: the arrangement, the automation, the reverb tail that resolves exactly where you wanted it.

And you have all the CPU in the world. If a plugin chain takes four seconds to render one second of audio, that is fine, because you are not in a hurry. You bounce it, you wait, you get a file.

The output of a DAW is one performance, rendered once, identical forever. That is a feature. It is also the whole problem.

An audio engine is a realtime system with no idea what happens next

Wwise, FMOD, Unity's built in audio, Unreal's MetaSounds. These run inside the game, on the player's machine, while the player is doing something you did not predict.

The engine wakes up every few milliseconds and answers one question: what sound comes out of the speakers right now. To answer it, it is doing work your DAW never has to do.

• Voice management. There is a hard cap on how many sounds can play at once. Forty enemies fire at the same time, and something has to decide which shots you actually hear and which get dropped. That decision is called voice stealing and it happens whether you planned for it or not.
• Memory and streaming. Short sounds get held in memory. Long ones get streamed off disk. Get that split wrong and you either blow the memory budget or hitch every time the music changes.
• Space. Distance attenuation, panning, occlusion when a wall is in the way, reverb that changes when the player walks into a cave. None of that is in your bounce. It cannot be. It depends on where the player is standing.
• A budget. All of the above has to finish inside the audio callback, which is on the order of five milliseconds. Miss it and the player does not hear a slightly late sound. They hear a crackle.

The output of an audio engine is a different performance every time anyone plays.

The part that catches most composers out

A game does not want your song. It wants a system that can produce your song, plus every other version of your song the game might need.

Combat starts. The boss drops below twenty percent. The player dies mid phrase. The music has to answer all of that, and a stereo bounce cannot answer anything.

So you do not ship a track. You ship parts and rules:

• Vertical layers. The same bar, written as separate stems. Tension rises, the engine fades in the strings and the percussion. Nothing restarts. The music got more intense without going anywhere.
• Horizontal segments. Chunks that the engine can jump between, but only on a musical boundary, so the transition lands on the beat instead of whenever the trigger happened to fire.
• Real loop points. Sample accurate, no click, no gap. Bounce a loop to mp3 and the encoder pads the ends with silence, which is exactly the click you are trying to avoid. Hand over wav and let the engine compress it per platform.
• Tails as one shots. When the music has to stop, something has to ring out. That decay is its own file, not the last two seconds of a loop you were hoping nobody would cut.

This means writing while knowing the music will be taken apart. Consistent tempo grid. Consistent key across anything that might need to crossfade. It is a real constraint, and it is the job.

Your mix is not the final mix

In a DAW you mix once and it stays mixed. In a game, dialogue ducks the score, an explosion squashes everything for 200 milliseconds, and the whole balance shifts depending on what is on screen. Those are runtime decisions, made by the engine, driven by game state.

Which means the loud, limited, ready for Spotify master is the wrong thing to hand over. It has no headroom left for the engine to work with, and you have already spent the dynamic range the game wanted to use.

Give it stems, mixed but not crushed. Consistent loudness across your cues so the engine is not fighting you. No master bus limiter. The engine gets the last word on the mix, and it is going to take it whether you left room or not.

The short version

A DAW: offline, fixed timeline, unlimited CPU, one output.

An audio engine: realtime, unknown timeline, hard CPU and memory budget, a different output every session.

They are not competitors and neither one replaces the other. The DAW is where the music gets written. The engine is where it gets used.

We are building both ends of this

The reason this seam is on our minds so much is that we are building both sides of it.

KOSMIK is our DAW. Zenith is our audio engine. Both are in development. Neither is finished, so there is nothing to buy at the bottom of this paragraph and no list to join yet.

Building one of those is a reasonable thing for a company like ours to do. Building both sounds like overreach, right up until you have written a score, handed it over, and watched it come apart in the gap between the tool that made it and the system that had to play it. That gap is the entire subject of this post. We would rather own both sides of it than keep writing articles about it.

When they are ready for you to actually get your hands on, they will get a page of their own. This is not that page.

In the meantime, hand us the whole chain

This seam is where game audio usually goes wrong, because it sits between two jobs. The composer delivers a beautiful wav and considers it done. The programmer receives a beautiful wav and has no idea it was supposed to have four layers.

Galactrex Records writes the music, and Galactrex Studios puts it in the game. Same building. That means you get stems, loop points, transition logic, and the implementation, instead of a folder of files and good luck.

If you have a game that needs a score that can actually react to it, talk to us. Bring the game. We will bring the parts that move.

Your DAW has all the time in the world to get it right. The audio engine has five milliseconds and no idea what the player is about to do.