Why captions go beyond words
Speech is only part of what a soundtrack says. A knock at the door, a sudden silence, laughter in the next room or a tense musical theme can all shape how a scene is understood. Hearing viewers pick these up without effort. A viewer who relies on captions can only receive them if someone writes them down.
This is the purpose of non-speech information: short text descriptions of sounds that matter to the story or the situation. Subtitles intended only for translation often leave them out, while captions written for deaf and hard of hearing viewers typically include them.
What counts as non-speech information
Several kinds of sound are commonly described.
- Sound effects. A phone ringing, a car crash, footsteps approaching, glass breaking.
- Vocal sounds that are not words. Laughter, crying, gasps, sighs, coughs.
- Music. Whether music is playing, its mood, and sometimes its title or lyrics.
- Ambient or background sounds. Rain, a crowd, a machine humming, if they add meaning.
- Silence or change in sound. A note that the music stops, or that a speaker is whispering, when it affects meaning.
- Off-screen speech. A note that a voice comes from offscreen or from a device such as a radio.
Deciding what to include
Not every noise needs a caption. Captioning every creak and rustle would clutter the screen and leave little time to read the dialogue. A useful test is whether the sound:
- moves the plot forward or explains an action,
- is a cue that characters react to,
- sets a mood the viewer would otherwise miss, or
- warns of something important, as with an alarm or siren.
If the answer is no, it is usually fine to leave it out. Style guides differ in how much detail they expect, and the right level may depend on the type of content. A drama may call for more mood description than a lecture, where the sounds of a classroom are rarely relevant.
Common ways to write them
Conventions vary, but several patterns are widely used:
- Brackets or parentheses around the description, for example “[door slams]” or “(phone ringing)”.
- Lowercase or sentence case for the description, to keep it visually different from speech.
- Present-tense verbs or short noun phrases rather than full sentences.
- A musical note symbol to mark lyrics or music, in places where the display supports it.
What matters is being consistent within a project, so viewers learn the pattern quickly. Following the rules of a chosen style guide, or the requirements of a particular platform, avoids a mix of styles.
Writing the descriptions well
Good descriptions are short, specific and neutral.
- Be specific. “[thunder rumbles]” tells more than “[noise]”.
- Say what, not how it feels, unless the feeling is part of the point. “[tense music]” can help set a scene, but over-interpreting can mislead.
- Do not give away a surprise early. If a sound is a reveal, the caption should appear when the sound happens, not before.
- Avoid repeating what is visible. If a character plainly slams a door on screen, the caption still helps a viewer who cannot hear, but repeated detail should be kept short.
- Use plain words. Avoid jargon or invented terms.
Handling music and lyrics
Music is often the hardest case. Several choices exist:
- A simple label, such as “[upbeat music]”, when music is only background.
- A title or artist where it is known and helpful, in cases where the content is meant to be identifiable.
- The full lyrics, when the words matter to the scene, such as a character singing or a song that comments on the story. Lyrics may involve copyright, so rights should be checked before including them. The wider topic of song lyrics is addressed in separate coverage.
- A note when music begins and ends, so viewers know the soundtrack has changed.
Foreign-language songs or speech raise another question, whether to translate, transcribe or simply label them, which is discussed under untranslated speech.
Timing and placement
A description should appear when the sound occurs and stay long enough to be read. Because sounds can overlap with speech, caption writers sometimes place the cue on a separate line or use a position away from the dialogue. Short cues need the same minimum reading time as any other text, which ties into reading speed.
Automatic captions and sound cues
Many automated captioning tools transcribe speech but do not describe sounds, or do so only in a limited way. Where accurate sound information is important to understanding, a person should add or check it. For safety-related audio, such as alarms or instructions, a qualified reviewer should confirm that the captions convey what a hearing person would understand.
A hypothetical example
Imagine a short scene in which a character is reading in a quiet house when a window shatters. A well-captioned version might show “[window shatters]” at the moment it happens, followed by the character’s gasp as “[gasps]”. It would not describe the ticking clock heard earlier, unless the clock was important to the scene. The aim is to give the viewer the same understanding as a hearing viewer would have, without crowding the screen.