Brief a Film's Audio Description as Part of Its Creative Plan
Brief a Film’s Audio Description as Part of Its Creative Plan
A useful audio description brief tells a describer three things about your film: which visual facts a viewer needs in order to follow it, when the film makes each of those facts available, and what the existing soundtrack is already telling everyone. Wording, voice and delivery choices come after those three, and depend on them.
That is the shape of the job. It is not a narration script, and it is not an accessibility claim. A brief written at the pitch or rough-cut stage is a set of priorities and open questions that a describer, an editor and a sound team can act on — and that you can put in front of blind and low-vision reviewers without having to explain what you meant.
The public reference most films can start from is the W3C Web Accessibility Initiative's guidance on description of visual information, which I'll draw on below. It is written for web media — players, transcripts, separate audio files — so it supports access considerations rather than settling what any particular film should do with its own two minutes. The parts that travel well are the distinctions it keeps making: information a viewer needs versus information already in the audio, and the timing and audibility of anything you add.
One listening pass, two lists
Before you write a single sentence of description, sit with one passage and make two lists.
The first is what is visible: the facts a stranger would need in order to follow the action, the relationships or the formal pattern. Not everything on screen — the relevant things. W3C's guidance frames description around visual information needed for understanding, which is a more useful test than completeness. The color of a coat is usually irrelevant. A hand moving toward a pocket is not.
The second list is what the main audio already delivers. Dialogue does obvious work. So do footsteps, a door, a score that tells you a scene has turned, a silence that lands like a dropped plate. Anything established by sound does not need description; describing it is how a track starts talking over the film it is meant to serve.
The two lists overlap more than most filmmakers expect, and where they overlap is where the brief gets interesting. A viewer who can hear the shot fired already knows a shot was fired. What they may not know is who fell.
An annotated passage
Here is an invented scene, roughly twenty seconds long, with invented characters. Nothing in it has been described, mixed or reviewed — I'm using it to show the shape of the brief, not to report a result.
A person sets a sealed envelope on a kitchen table. The address side isn't readable. A latch clicks offscreen — no dialogue, just the sound. The person turns toward it, picks the envelope up and slides it under a magazine. Another person enters and speaks. Some beats later, a phone on the table lights up and an onscreen message identifies the sender of the envelope as the first person's estranged daughter.
The cue sheet might look like this.
| Beat | Visible | Already audible | What the audience is meant to know | Description candidate |
|---|---|---|---|---|
| 1 | A sealed envelope is set on the table; no address readable | Room tone | An envelope has arrived. Its sender is unknown | "A sealed envelope on the kitchen table." |
| 2 | Nothing new | A latch clicks offscreen | Someone is arriving | None. This is the sound's job |
| 3 | The person turns, picks up the envelope, slides it under a magazine | The tail of the click, a chair scrape, footsteps leaving frame | The envelope is being hidden. The sender is still unknown | "They slide it under a magazine." |
| 4 | A second person enters | Offscreen dialogue | The concealment had a reason | Contested — see below |
| 5 | A phone lights; the message on screen names the sender | Notification tone | Who sent it | "A message lights the screen. It names the sender." The onscreen text has to be voiced, not summarised |
Two lines in that table are doing real work, in opposite directions. Beat 2 requires no description, because the click is the information. Beat 5 requires description of a particular kind: the sender's identity arrives as onscreen text, and a blind viewer cannot read it, so the describer has to speak what the screen says. The film still controls the reveal — it happens at the phone and not before — and the brief's job is to say both things at once. A brief that does is doing more for the describer than any amount of style guidance.
And notice what the cue sheet refuses to state: whose handwriting is on the envelope, whether the person recognizes it, what the second person wants. Those are the film's open questions, and the brief's job is to protect them from a description that would rather have a tidier sentence.
The sentence that spends the film's reveal
The clearest way to brief a reveal is to write the version that destroys it and say why.
"She sets down a letter from her estranged daughter." Three problems in ten words. It names the sender before the phone message does, so the reveal at beat 5 has nothing left to show. It asserts the relationship — estranged — which the audience hasn't been told yet and may be learning gradually. And it settles the envelope's meaning before the concealment has had a chance to mean anything.
"She hides the envelope, afraid of what her daughter might want." This one is a different species of error. It doesn't just arrive early; it supplies a motive the scene leaves open. Fear is one possible reading of the concealment. So is irritation, or habit, or not wanting to explain something to a guest. A described version that picks one reading has narrowed the film for some viewers and not others — which is the opposite of what description is for.
"She hides it from the person at the door." Closer, and still too certain. The scene shows a turn toward an offscreen sound. That is compatible with knowing exactly who is arriving and with having no idea. Description doesn't have to resolve that. "They turn toward the sound and slide the envelope under a magazine" leaves the audience where the picture leaves them.
That progression is the whole craft of the brief in miniature. The first sentence over-names. The second over-motivates. The third over-infers. Each is more accurate than the last, and the last is accurate because it stops at what the image actually shows.
Sound occupies time, including the quiet
The hardest part of the brief is not wording. It's arithmetic.
In the invented scene, beat 3 is where the concealment happens, and it sits directly against the latch and the start of the incoming voice. A gap in dialogue is not a gap in the soundtrack. The tail of the click, a chair scrape and a receding footstep are all doing narrative work in those two seconds, and they're doing it for exactly the viewer the description is for. If description covers the click, the listener loses the cue that explains why the person hides the envelope in the first place. The whole beat unravels from there.
So the brief should treat meaningful sound the way it treats reveals: mark it. Dialogue, music, silence, effects that carry story. Then name the moments where added speech would cost the film something.
W3C's web-media guidance connects the delivery of description to the time available and to the playback context, and notes that where a video has little or no space in the audio, a separate described version may be the route. That's a useful prompt, not a schedule. What it means for your film is a production decision you and the describer make together, informed by the actual mix — which is why the brief should ask the question instead of answering it.
Three honest options for beat 3:
- Late. Describe the concealment in the first clear beat after the voice, accepting that a listener learns about it a second or two after it happens. Sometimes nothing is lost; sometimes the next shot has already moved on.
- Partial. Establish the envelope in beat 1, where there is room tone, and let the concealment be inferred from the sound of the latch plus whatever the next beat reveals. Less information, no collision.
- A different described version. More description, delivered at a pace that gives it room, with the film's own sound handled to suit. This is a different listening experience, and it may not be available in every place the film plays.
Each has a cost, and the brief should say which cost it is choosing. What it should not do is instruct a describer to speak faster across the latch.
Brief the wording; don't ghost-write the describer
You know things the describer can't get from one viewing, and those are the things worth writing down: recurring terms the film uses for its own objects, visual motifs that return, names the film deliberately withholds, and naming questions you haven't resolved yet. If a character's name is spoken in act two and not before, say so. If a room's layout matters because a later scene depends on it, say so.
W3C's web-video guidance recommends objective wording and a neutral voice, and it's worth naming that scope precisely, because filmmakers sometimes hear it as a rule about cinema. It isn't. It's guidance for web media, and it's a sound default: description that reports rather than editorializes keeps the film's judgment in the film. If your work wants something else — a description in a particular register, a narrator who is a presence rather than a reporter — that is a conversation to have with the describer and with reviewers. What it can't be is an assertion that tone outranks comprehension, settled in a brief before anyone has heard the result.
Plan who listens, and to which version
A brief should name the version of the film it describes. Picture lock, which mix, which delivery. Timing decisions made against one cut are void against the next, and a brief that doesn't say which cut it describes will be quietly wrong within a week.
Then name the review. Who evaluates the described version, and specifically: are blind and low-vision viewers among them? Two questions are worth putting in writing for whoever listens:
- At each reveal, is the information the viewer needs available at a moment when it's useful — not earlier, not so late that the next beat has passed?
- Is the sound the film depends on still perceptible under the description?
Listening to your own film without looking is a genuinely useful exercise, and it will surface questions you didn't know you had. It is not audience review, and the answers it produces are hypotheses. The presence of a description track on a file is also not evidence that the track works. Those are separate claims, and only one of them can be made from a script.
What the brief leaves open
The end of a good brief is not a summary. It's the annotated passage, the questions assigned to specific people, and at least one problem you haven't solved.
For the invented scene, that unsolved problem is beat 3: a concealment that matters, sitting on top of a sound cue that matters, with a voice arriving immediately after. The brief can say that the latch must remain audible, that the sender's identity must arrive when the phone's message does and not before, and that the motive is not the description's to supply. It cannot yet say where the sentence about the magazine goes, because that depends on a mix nobody has listened to with description in it.
Write that down as a question. A brief that hands collaborators a clear problem is more useful than one that hands them a tidy answer the film can't support — and it keeps the described version where it belongs: in the same creative conversation as the cut, the mix and the release.
Frequently asked questions
What three things should an audio description brief tell a describer before wording and delivery are chosen?
It should say which visual facts a viewer needs to follow the film, when the film makes each of those facts available, and what the existing soundtrack is already telling everyone. Wording, voice, and delivery choices come after those three and depend on them.
In the invented cue sheet, why is beat 2 undescribed, and why must beat 5 be voiced rather than summarized?
Beat 2 requires no description because the offscreen latch click already tells the audience someone is arriving. Beat 5 requires description of a particular kind: the sender's identity arrives as onscreen text, and a blind viewer cannot read it, so the describer has to speak what the screen says. The reveal still happens at the phone and not before.
What errors does the article use the envelope sentences to illustrate?
'She sets down a letter from her estranged daughter' names the sender before the phone message, asserts the estranged relationship early, and settles the envelope's meaning. 'She hides the envelope, afraid of what her daughter might want' supplies a motive the scene leaves open. 'She hides it from the person at the door' over-infers from a turn toward an offscreen sound. The better description stops at what the image shows.
Why is beat 3 difficult to brief, and what options does the article list?
Beat 3 is where the concealment happens and it sits directly against the latch, a chair scrape, a receding footstep, and the start of incoming voice. A gap in dialogue is not a gap in the soundtrack. The options are late description after the voice, partial description that establishes the envelope earlier and lets the concealment be inferred, or a different described version with more room. Each has a cost, and the brief should say which cost it is choosing.
What should be at the end of an audio description brief instead of a summary?
It should include the annotated passage, questions assigned to specific people, and at least one problem not yet solved. For the invented scene, that unsolved problem is beat 3: the brief can require the latch to remain audible, the sender's identity to arrive when the phone message does, and the motive not to be supplied, but it cannot yet say where the sentence about the magazine goes because that depends on a mix nobody has listened to with description in it.