Design Captions, Transcripts, and Audio Alternatives
Accessibility · accessibility, microcopy
Updated 2026-07-28
- Assume most people watch with the sound off. Captions are an accessibility requirement and also the default viewing mode in feeds, offices, and public spaces.
- Caption meaning, not just words. Speaker changes, laughter, a door closing, and a long silence all change how a scene reads.
- Give captions a readable presentation:
- Two lines at a time, not a wall of text.
- Enough contrast against any frame behind them.
- A position that never covers on-screen text or faces.
- A size the user can change.
- Provide a transcript as well as captions. A transcript is searchable, skimmable, translatable, and usable by people who cannot follow moving text.
- Describe what only the picture shows. If a demo says "click here" while the cursor moves, the audio alone is useless, so either narrate the action or supply a described version.
- Never autoplay audio. Sound that starts on its own is disorienting for screen reader users, because it competes with the voice they are already listening to.
- Treat machine-generated captions as a first draft. Product names, jargon, and numbers are exactly the parts that matter and exactly the parts that come out wrong.
Media is only finished when it works with the sound off, the screen off, and the playback speed changed.
Related guides
UX Handbook