Subtitles are timed text displayed over or below the picture that renders the spoken dialogue of a video, most often translated into another language. Each subtitle event has an in-time, an out-time and one or two lines of text, synchronized so that it appears while the corresponding speech is heard.
In video accessibility, subtitles sit next to closed captions, open captions and audio description. Classic subtitles assume the viewer can hear the soundtrack but not understand the language, while subtitles for the deaf and hard of hearing (SDH) add speaker identification and relevant sounds. In British usage the word subtitles often covers captions as well, so the exact meaning depends on market and context.
How subtitles are created and synchronized
Subtitles are produced in a pipeline of transcription, translation, segmentation and timing. The source dialogue is transcribed, translated, condensed to fit reading speed, split into readable events, and spotted against the picture so each event starts and ends at sensible moments.
- Template creation: a source-language file with timing and line breaks serves as the master for all languages.
- Translation and condensation: translators shorten speech to fit reading speed limits while keeping meaning.
- Spotting: in-times and out-times are set to speech onsets and adjusted to shot changes.
- Quality control: checks cover timing overlaps, line length, reading speed, spelling and on-screen text conflicts.
- Delivery: the file is exported to the format required by the broadcaster, platform or disc authoring system.
Timing is expressed in timecode or milliseconds, and it must match the frame rate of the delivered video. A file timed at 25 fps and played against a 23.976 fps master drifts progressively out of sync, which is one of the most common delivery errors.
Types of subtitles and how they differ from captions
Subtitles differ by audience, content and rendering method. The table below summarizes the main variants and places them next to the neighboring Terms of the Topic, which have their own detailed pages.
| Variant | Intended viewer | What it contains | Can be turned off |
|---|---|---|---|
| Interlingual subtitles | Hearing viewer, foreign language | Translated dialogue | Usually yes |
| Intralingual subtitles | Hearing viewer, same language | Transcribed dialogue | Usually yes |
| SDH | Deaf or hard of hearing viewer | Dialogue, speaker IDs, sound cues | Usually yes |
| Forced narratives | All viewers of a language version | Foreign speech and on-screen text only | No, shown automatically |
| Closed captions | Deaf or hard of hearing viewer | Dialogue and sound information | Yes |
| Open captions | All viewers | Text burned into the picture | No |
The decisive distinction is what the text assumes about hearing. Standard subtitles omit sounds such as a doorbell or music cues, because the viewer is expected to hear them. That makes standard subtitles insufficient as an accessibility feature for deaf viewers, even when they share the same language as the audio.
Subtitle file formats and delivery
Subtitles are delivered as text-based or image-based tracks. Text formats are small, searchable and restyled by the player, while image formats preserve exact appearance but cannot be resized or restyled by the viewer.
| Format | Type | Typical use | Styling support |
|---|---|---|---|
| SubRip (.srt) | Plain text | General web and desktop playback | Minimal |
| WebVTT | Text | HTML5 players and HLS streaming | Positioning and basic styling |
| TTML and IMSC | XML text | Broadcast, OTT platforms, MPEG-DASH | Rich, profiled for interoperability |
| EBU STL | Binary text | European broadcast archives and exchange | Teletext-style |
| DVB and PGS subtitles | Bitmap images | Digital television and Blu-ray | Fixed by authoring |
For streaming, subtitles travel as a separate selectable track rather than inside the picture, which lets one video master serve dozens of languages. Burned-in subtitles remain common on social media, where autoplay is muted and many players ignore sidecar tracks.
Readability rules and timing parameters
Readable subtitles follow limits on line length, reading speed and display duration. Exact values vary by broadcaster and language, but most professional style guides converge on similar ranges for Latin-script languages.
| Parameter | Common range | Why it matters |
|---|---|---|
| Characters per line | 37-42 | Keeps lines within comfortable eye span |
| Lines per event | Maximum 2 | Limits how much of the picture is covered |
| Reading speed | About 15-20 characters per second | Viewers must read and still watch the image |
| Minimum duration | About 0.8-1 s | Very short events cannot be read |
| Maximum duration | About 6-7 s | Long events cause viewers to reread |
| Gap between events | About 2 frames | Signals the change to a new subtitle |
Good subtitlers also break lines at linguistic units, keeping articles with nouns and verbs with their objects, and they avoid carrying a subtitle across a shot change, which causes the eye to reread the text. Children’s content uses lower reading speeds.
Common subtitle errors and accessibility pitfalls
Subtitle problems usually come from timing drift, overloaded events or poor placement. Each one reduces comprehension and, for viewers who rely on text, can make content unusable.
- Frame rate mismatch causes slowly increasing desynchronization over the running time.
- Overlapping on-screen text: subtitles cover lower thirds or signs unless they are repositioned to the top of the frame.
- Low contrast: white text without an outline or background box disappears over bright scenes.
- Unedited automatic transcripts contain misheard words, missing punctuation and no speaker changes.
- Treating subtitles as captions: omitting sound cues fails deaf viewers and accessibility requirements such as WCAG success criteria for captions.
The idea in brief
Subtitles are the lines of text at the bottom of the screen that show what people are saying. Most often they translate a film made in another language.
They appear and disappear in step with the speech. They are kept short so you can read them and still watch the picture. Regular subtitles assume you can hear the music and sounds. Versions for deaf and hard of hearing viewers also describe those sounds.
A practical scenario
A documentary producer prepares a 52-minute Spanish-language film for release on an international streaming platform in 12 languages. The subtitle vendor first builds an English template with 640 events, line length capped at 42 characters and reading speed at 17 characters per second, then translates it into each target language.
Quality control finds that the German file was timed against an early 25 fps cut, while the final master runs at 23.976 fps, so the subtitles drift by more than two minutes by the end credits. After retiming to the final master and adding forced narratives for on-screen street signs, all files are delivered in IMSC, and every language version plays in sync on launch day.