What does your AI film say to someone who cannot see the picture?
Subtitles explain speech and useful sounds; they do not automatically describe a silent action. I would listen to the film without looking at it, identify which important information exists only in the picture and plan how to make that information available. The right answer may involve the script, a described version or a descriptive transcript.
The missing action may be the whole story
Imagine a fictional workshop film: a visitor tries two handles, chooses one and leaves with a finished object. The audio contains music and “Made for your day”. A listener may hear the mood but miss the decision. If choosing the handle explains the offer, that action matters more than the colour of a wall or the shape of an unimportant background object.
I would list the meaning that disappears, not try to narrate everything visible. Who acted, what changed and what does the viewer now know? The answer keeps description connected to the film’s purpose. It also helps avoid filling every quiet moment with detail that competes with dialogue or prevents the audience from enjoying the sound.
Separate sound information from visual information
Start a scene sheet with three columns: what the picture shows, what the existing audio communicates and what remains essential but missing. A subtitle file can include speech and relevant sounds, yet still leave a visual choice unexplained. Likewise, a written description beside the player may help some readers without providing the same viewing experience as an integrated described track.
The worksheet asks you to make that distinction for one scene. Its result is a draft review note. It cannot listen to your film, decide which details are essential or certify accessibility. A person has to compare the actual picture and sound, and people who use the intended access format should be involved in review.
| Material | Carries | Check carefully |
|---|---|---|
| Captions | Speech and relevant sound information | Timing, speakers and readability |
| Audio description | Essential visual meaning in sound | Accuracy and space around existing audio |
| Descriptive transcript | Words, sounds and visual information in text | A clear route to the right version |
One scene, three passes
Use the buttons or the left and right arrow keys to follow the sequence.
Scene 1 / 3
Picture
A visitor chooses one handle from two.
Sometimes the original script can do the work
If the scene already has a narrator, a precise line can make the action understandable without adding a second voice. “Choose the handle that feels right in your hand” gives the listener more than a general brand slogan. It still needs to fit what really happens. A line should not say that a person selected a custom option if the picture only shows an ordinary finished object.
Planning this before the edit can leave room for information instead of squeezing it into the final seconds. For a wordless film, another described version or a descriptive transcript may be worth planning. Which materials are required depends on the content and the applicable context; this article is a production worksheet, not a compliance certificate.
Find the meaning that only the picture carries
Use this local worksheet with your own project. The starting values are an illustrative example.
Your working note
This worksheet runs in your browser. It sends no form entries to the server and does not save them after reload. Download a copy before leaving.
Complete every text field and keep numbers inside the stated limits.
Review the timing as carefully as the words
Description needs a place to live. Read the draft against the actual sequence and check whether it clashes with dialogue or covers an important sound. If there is no usable pause, discuss another editorial solution rather than speaking faster until it fits. The listener needs the meaning at a useful moment, especially when a later action depends on what happened before.
Keep the version relationship clear. An updated edit can move a choice, remove a person or change the closing instruction; the description then needs another review. Store the approved text with its matching film version and identify the final playback route. A good document that nobody can find beside the player does not solve access for the visitor.
Make the access route part of the delivery
Ask someone to reach the described material from the actual page using the intended controls. Check the names of the versions, the text link and the ability to pause or stop playback. Then ask what remained unclear. That is more useful than ticking a generic “accessible” box after uploading a file. Keep specific findings and fix them in the corresponding version.
For SOLONSKY Films, I would include these needs in the brief before sound and editing are locked. Bring the intended audience, destination and available review support. We can identify the needed materials and agree who checks them. Accessibility is a sequence of concrete production decisions, and a visual story benefits when those decisions are given room from the beginning.
- Identify essential visual-only meaning
- Compare it with the real soundtrack
- Choose a suitable description route
- Review timing with the matching edit
- Test how someone finds the finished material
Listen first, then compare with the picture
Try The Impossible Brief as an observation exercise: listen without looking, then watch. This is not a claim that the linked film has an audio-description track or has passed an accessibility audit.
The video provider connects only when you press play. Watch on YouTube.
Questions
Do captions replace audio description?
They serve different information needs. Captions alone may not communicate essential visual actions that are absent from the soundtrack.