Tutorials Audio

Scoring for Picture: A Composer's Workflow

Intermediate · ~30 min

Overview

Scoring is composition constrained by picture, and the constraints are what make it a distinct craft. The music has to land on specific frames, support a scene without competing with dialogue, survive being remixed on a dub stage, and change when the edit changes, which it will, repeatedly, after the music is written. This guide covers the workflow from spotting session to stem delivery, the technical mechanics of sync, and the practical realities of writing music that has to share a soundtrack with dialogue and effects.

What You Need

A DAW that handles video playback and timecode. Most major ones do
A picture lock or, realistically, a working cut you accept will change
A sample library or instruments appropriate to the palette, or players to record
Monitoring you trust, plus a check on small speakers
A cue sheet or spotting notes agreed with the director and editor
Enough disk and session organisation to manage many revisions without losing versions

Steps

Spot the film before writing a note

A spotting session is where you and the director decide where music starts, where it stops, and what each cue is doing. The decisions about where music should be absent matter as much as where it plays, wall-to-wall scoring flattens a film and removes the contrast that makes entries land. Record the agreed in and out points as timecode, and name each cue systematically.

Establish tempo against the picture, not in the abstract

The music has to hit specific moments, so tempo is a tool for alignment rather than a starting preference. Find a tempo where the important frames fall on musically sensible beats. Where they will not align, use a tempo map with gradual changes rather than forcing the picture to accommodate a fixed grid, audiences notice music arriving late on a cut far more than they notice a tempo drifting.

Write themes that survive transformation

A theme is useful in a score when it can be recognised while being changed, slowed, reharmonised, reduced to a fragment, played by a different instrument, turned from major to minor. Write material simple and distinctive enough to remain identifiable through those transformations, because that is how a score carries meaning across a film rather than just accompanying individual scenes.

Orchestrate around the dialogue, not over it

Speech occupies the frequency region where most melodic instruments also sit, so music competing there forces the mixer to bury it. Write with the finished soundtrack in mind: keep sustained mid-range material sparse under dialogue, use lower and higher registers for presence, and place your most active writing in the gaps between lines. This is the single biggest difference between music that survives the mix and music that gets turned down.

Hit the important moments precisely, and let the rest breathe

Identify the few frames in a cue that genuinely require synchronisation (a door opening, a realisation, a cut to a new location) and place musical events exactly there. Attempting to catch every action produces the frantic, cartoonish effect known as mickey-mousing. Precision on a small number of moments reads as intentional. Precision on everything reads as comic.

Expect the edit to change and structure sessions accordingly

Picture lock is frequently provisional. Build sessions so that cues can be shortened, extended, or shifted without rewriting: keep material modular, avoid burying tempo changes in ways that are painful to move, and retain the flexibility to re-cut. Composers who assume the picture is final spend their revision time rebuilding rather than improving.

Mix the cue so it works within a soundtrack

A cue that sounds impressive in isolation is often wrong for the film, because it will be sharing space with dialogue and effects. Check your cues against the production sound wherever possible, and on small speakers as well as good ones. The question is not whether the music sounds good alone but whether it does its job when everything else is present.

Deliver stems, documentation, and correct paperwork

Dub stages need the score split into stems (commonly strings, brass, woodwind, percussion, synth, and any solo or vocal element) so the re-recording mixer can rebalance under dialogue without asking you to re-render. Deliver at the agreed sample rate and bit depth, starting at a consistent timecode, named to the cue list, with a cue sheet documenting timings and authorship for rights registration.

Pro Tips

Silence is a scoring choice. The most effective entry in a film is often preceded by several minutes of nothing.
Watch scenes without music repeatedly before writing. The scene tells you what it needs more reliably than a brief does.
Name and number cues systematically from the start, reel and position, or revision management becomes chaos by the third pass.
Keep a rendered reference of every approved version. Directors regularly ask to return to something from two revisions ago.
Print stems as you go rather than at the end. Stem delivery under deadline is where avoidable disasters happen.
If the director says a cue is "too much", the problem is usually density in the mid-range, not volume.

Knowledge Base

What You'll Learn

This page is the craft entry point for scoring, a discipline this site previously covered only through composer biographies. Below: what spotting decides, how sync actually works, why orchestration for film differs from concert writing, and what delivery involves.

What a score is actually doing

Film music performs several distinct jobs, and confusion about which one a cue is doing produces most of the notes a director gives.

It can establish emotional context, telling the audience how to read an ambiguous image. It can maintain continuity, binding a sequence of disparate shots into one experience, one of the most common and least noticed functions. It can control pace, making a slow sequence feel purposeful or a fast one feel urgent. It can carry information through thematic association, so a returning motif tells the audience something the image does not.

And it can work against the picture deliberately (cheerful music over violence, stillness under chaos) which is among the most powerful tools available and the easiest to overuse.

Naming the job of each cue during spotting makes revisions far more productive, because "this is not working" becomes a discussion about whether the cue is doing the agreed job rather than a matter of taste.

Tempo, sync, and hitting the frame

Synchronisation is the technical core of scoring, and it comes down to reconciling musical time with picture time.

Picture time is measured in timecode, hours, minutes, seconds, frames. Musical time is measured in bars and beats at a tempo. A hit point is a frame where something must happen musically, and the craft is arranging the tempo so that those frames fall at musically defensible places.

Where a single tempo cannot catch every hit, the standard approach is a tempo map: gradual changes across a cue that shift the grid so the important moments align. Done well this is inaudible, because audiences accept tempo movement in expressive music. Done abruptly it sounds like a mistake.

Not every hit deserves precision. Deciding which three or four moments in a cue genuinely require frame accuracy, and letting everything else float, is what separates a score that feels composed from one that feels mechanically synchronised.

Orchestrating for a soundtrack rather than a concert hall

The defining constraint of film scoring is that the music is not the only thing playing. It shares a soundtrack with dialogue, which has absolute priority, and effects, which frequently have priority too.

Dialogue intelligibility depends heavily on a mid-range band where a great many instruments also live. Sustained material in that region under speech forces the mixer to reduce the music, which is why cues that seemed full in isolation can end up nearly inaudible in the finished film.

The practical responses are consistent across the craft: keep the mid-range sparse under dialogue and use registers above and below it for presence. Favour movement and rhythm over sustained harmony when speech is dense. Place the most active or melodic writing in the gaps between lines. And treat the spaces where nobody is speaking as the opportunities they are.

Composers who write with the mix in mind get more of their music heard. This is not a compromise of the writing. It is the actual discipline.

Temp music, and the trap it sets

Editors cut with temporary music, existing recordings dropped in to give a sequence shape during editing. It is genuinely useful and creates a well-known problem.

Directors watch the cut with temp music dozens of times and become attached to it. The brief that follows is then frequently a request to reproduce the temp closely enough to feel familiar while being different enough not to infringe. An uncomfortable position that produces derivative work and occasional legal exposure.

The workable approaches are to be involved early enough to influence what gets used as temp, to discuss explicitly what the temp is doing rather than what it sounds like, and to redirect conversations from "make it like this" toward the function the temp is serving. The last of these is the most useful skill, because the underlying request is nearly always about function even when it is expressed as sound.

Delivery: stems, cue sheets, and rights

Delivery is where scoring meets the production pipeline, and getting it wrong creates work for other departments under deadline pressure.

Stems are the score split into instrument groups, rendered separately at consistent start times so they recombine exactly. The re-recording mixer uses them to rebalance the score against dialogue and effects, lowering strings under a line without losing the percussion, for instance. Delivering only a stereo mix removes that flexibility and guarantees a request for revisions.

Cue sheets document every piece of music: title, composer, publisher, duration, and usage type. They are the basis for performance royalty distribution, so an inaccurate or missing cue sheet is a direct financial loss to everyone credited.

Rights are worth settling in the contract rather than at delivery. Who owns the score, who publishes it, whether a soundtrack release is permitted, and what happens to reuse in trailers and sequels are all far easier to agree before the work than after.

FAQ

Q: Do I need to read music to score?
A: Not to work in a DAW with samples and synths, which is how a large share of scoring now happens. You do need it to work with live players, since parts must be notated, and it makes communication with orchestrators and musicians vastly more efficient. Many working composers write in a DAW and collaborate with an orchestrator for notated material.

Q: How is scoring different from writing a song?
A: A song is self-contained and controls its own structure and duration. A cue is subordinate to picture: its length, its emotional arc, and its moments of emphasis are all determined by the scene, and it has to share a soundtrack with dialogue that takes priority. The compositional skills overlap. The constraints do not.

Q: What are stems, and why does the mixer need them?
A: Stems are the score rendered as separate instrument-group files that sum back to the full mix. They let the re-recording mixer rebalance the score against dialogue and effects, reducing sustained strings under a line while keeping rhythm intact, without asking you to re-render. Delivering only a stereo mix removes that option entirely.

Q: How do I handle a director who loves the temp music?
A: Move the conversation from sound toward function: ask what the temp is doing for the scene rather than which parts they like. Attachment to temp is nearly always about how it makes the sequence work, not about the specific recording, and articulating that gives you room to solve the same problem differently. Being involved before temp is chosen is the better prevention.

Q: When should music not play?
A: More often than most first scores assume. Continuous music removes contrast and denies you the impact of an entry. Silence before a significant cue, and dialogue scenes left unscored so they play on performance alone, are both standard tools. The decision about where music is absent is a genuine part of the spotting session.

Q: What sample rate and format should I deliver?
A: Whatever the post production supervisor specifies, and ask early rather than assuming, 48 kHz at 24-bit is the common baseline for picture, with higher rates on some productions. Deliver stems starting at a consistent timecode, named to match the cue list, and confirm the naming convention before you render a whole score to the wrong one.