A dialogue-heavy scene does not need to be score-free. It needs a score with a defined job. If performance and editorial rhythm already carry the information, music should not paraphrase every line. It can hold continuity, shift point of view or prepare a change the characters have not named yet.
Before I write, I ask what the scene would genuinely lose without music. That answer is more useful than a request for additional emotion. It also gives the director, editor and composer a shared test for the first mockup.
Separate information from dramatic movement
Dialogue can deliver plot, hide intention or fill the surface while a relationship changes underneath. Those are different scoring problems.
An exposition-heavy exchange may need continuity rather than commentary. A lie may need music that stays with the film’s perspective without exposing the speaker too early. A casual conversation that ends in separation may benefit from a slow musical trajectory that the individual lines never state.
I try to write the cue’s function as one sentence. “Keep the relationship fragile while the language stays playful” is actionable. “Make it more emotional” is not. A functional sentence also makes notes less subjective because everyone can ask whether the cue is doing the agreed job.
Spot the words that need clear air
Spotting dialogue is not only about an in point and an out point. It is about identifying words, breaths and looks that music must not crowd.
An entrance before a line can prepare it. The same entrance on the first word may underline it. Dropping out before a reveal can return focus to performance. Stopping exactly on the reveal may feel like editorial punctuation.
I therefore map the scene’s breathing as well as its timecode. Where does speech accelerate? Where does a pause become meaningful? When does the cut move us closer to a character? Music can then sit between decisions instead of landing on every decision.
Restraint is not the same as neutrality
Music under dialogue can be sparse and still specific. Register, texture and harmonic motion may establish identity without competing for attention. The issue is not simply note count. It is how many sources of information the audience is being asked to follow.
An active melody naturally pulls focus. Detailed rhythm may fight speech cadence. A dense low-mid arrangement can leave little room at the premix. I can move expression into timbre, unstable sustain or a gradual change that remains present without asking the audience to look away from the actors.
That restraint must still belong to the film. An interchangeable pad under the entire scene is not automatically sensitive. It fills space without explaining why music entered it.
Review mockups against the actual dialogue
I send dialogue cues against picture with the relevant production sound. Listening to the music in isolation creates the wrong density test. A cue that feels incomplete alone may be exactly complete once voices, room tone and effects return.
The review should answer three questions. Are the important lines clear? Does the performance keep its changes of weight? Does music add a reading that was not already present?
The mockup is not the final mix, but its balance needs to be credible. Hiding the cue does not prove that it works. Pushing it far above its likely dub level creates a problem the final mix was never meant to solve. I prefer a realistic presentation and a clear note about what remains flexible.
Give notes on consequence before solution
The most useful director notes describe what the cue does to the scene. “We think he is leaving too early,” “she now sounds more certain than the performance” or “the silence before the final line has disappeared” gives me a dramatic target.
“Use less piano” may be a valid instinct, but it begins with the proposed fix. The actual issue could be register, rhythm, timing or voicing. When the production describes consequence first, it keeps authority over the story while leaving the composer room to find the smallest musical correction.
Notes from the director, editor, producer and studio should be consolidated. A subtle change under dialogue can alter several lines at once. Independent feedback chains quickly produce a cue with no stable point of view.
Write with the dub in mind
The dub stage cannot turn a crowded arrangement into a clear dramatic idea without giving something up. I consider vocal range, dynamic movement and likely sound-design priorities while composing.
That does not mean banning the midrange. An instrument close to speech can create striking intimacy when it occupies the gaps. If it plays continuously, the re-recording mixer will need to make more compromises.
Stems provide useful control when they reflect the arrangement. Depending on the score, I may separate pulse, low end, harmonic texture, melodic material, soloists or orchestral families. The mixer can then clear a line without pulling the entire score down. Stem architecture belongs in the delivery conversation early, not in a panicked export session before the dub.
Recheck meaning after picture changes
Dialogue scenes rarely survive blind reconforming. Removing a few frames can move a breath, place a harmonic change under a different word or erase the pause that justified the cue’s entrance.
When a new picture version arrives, I check function before mechanics. Is the scene still making the same argument? If it is, a musical edit may be enough. If character alignment has shifted, a technically accurate reconform may preserve duration while losing meaning.
A change list and a clearly named reference export let me separate conform work from rewriting. They also help production understand why some revisions are quick and others affect the cue’s structure.
No score is still a scoring decision
Some conversations work better without music. That is not a failure to compose. Unscored space can let the audience judge a line, hear a room or sit with discomfort without being told how to resolve it.
I would rather make that choice during spotting than discover it after several revisions. The absence then becomes part of the wider music plan. A later entrance carries more weight because the previous scene was allowed to stand alone.
Good music under dialogue does not need to advertise itself. It protects the work already present in performance and editorial, then adds a trajectory the words cannot carry by themselves. That requires a functional brief, picture-based review, consolidated notes and deliverables designed for the dub. It is quiet craft, but it is often where a score earns the production’s trust.