Diegetic means the characters can hear it. Non-diegetic means only the audience can. It sounds like a term you learn for an exam, but it is a working decision that changes how you treat every sound in the scene.
The definitions, then the part that actually affects your mix.
Diegetic sound exists in the world of the story. A car passing, a door, a line of dialogue, a radio playing in the kitchen. If a character could turn round and look at the source, it is diegetic.
Non-diegetic sound is added for the audience alone. The score, a narrator who is not in the scene, a whoosh on a transition. The characters cannot hear any of it.
Source music is the useful term in between, and it is just diegetic music: music with a visible or implied origin in the scene. A band in the bar, a car stereo, a party upstairs.
The word diegetic comes from diegesis, the Greek word for the narrated world. That is the whole etymology and it is genuinely the clearest way to remember it: inside the world, or outside it.
This is the part the film-studies version leaves out. The label is not academic, it is a commitment to a set of rules.
If music is playing from a radio across the room, it has to sound like it. It gets the room's reverb, it gets filtered by whatever is between it and the camera, it pans with the radio, and it changes when the camera moves. That is real work, and it is what the placement guide is about.
Score sits wherever you put it. It does not need a position, it does not get filtered by walls, it does not change when you cut to a different angle. It is free of the physics, which is exactly why it can carry emotion the scene cannot.
Deciding late is expensive. If you mix a track as score for a week and then decide it is coming from the jukebox, every one of those choices has to be redone with the room in it. Make the call before you start treating the cue.
Ask whether a character could plausibly turn it off. If yes, it is diegetic and it obeys the room. If the idea of turning it off is nonsense, it is score and it floats free.
The interesting work happens when you break the rule deliberately. There is a term for sound that sits on the boundary, trans-diegetic, and the moves are worth knowing because they are cheap and they land hard.
The technical version of the first two is a transition between a heavily treated version of a track and an untreated one. Done as a hard switch it is jarring. Done as a gradual opening of the filter, a rise in level and a widening of the stereo image over two or three seconds, it is invisible and it works.
Once you have decided a cue is diegetic, the job is placing it, and placing it convincingly is mostly about the room between the source and the camera rather than about the music itself. The filter settings for walls, doors and floors are here, and if the sound is coming from a device rather than through architecture, the futz numbers are here.
That trans-diegetic transition, opening a track up from a room into the score, is three parameters moving in sync over a few seconds, which normally means drawing three automation lanes and redrawing them whenever the edit changes.
THALM makes it one move: arm the record mode, press play, and drag the listening position out of the building while the scene runs. The filtering, the level and the space open together because the geometry changed, and what you get back is an editable route rather than a print.
We write these when there is something worth writing down. One email when a new one lands or a new Tunary instrument ships. No newsletter, no schedule.
Tunary