Subtitles, captions, and why the timing drifts
Subtitles and closed captions are not the same feature, and when timing drifts out the cause is usually a frame rate mismatch rather than a badly made file.
Two things get called subtitles and they serve different audiences. The distinction matters when you're choosing a track, and it explains why one option sometimes has far more text than another.
Subtitles versus captions
Subtitles assume you can hear the audio but not understand it — typically a translation, or dialogue in a language you don't speak. They transcribe speech and nothing else.
Closed captions assume you can't hear the audio. Alongside dialogue they describe relevant sound: a door slamming, music swelling, who is speaking when it's ambiguous. Captions are an accessibility feature with a legal basis in many countries, which is why broadcast content reliably has them.
Menus often label both as "subtitles", with captions marked "SDH" — subtitles for the deaf and hard of hearing. If you want sound description, pick SDH. If you just want translated dialogue without a running commentary on the score, pick the plain track.
Why sync drifts
Almost always a frame rate mismatch, and the pattern tells you which kind.
Gradual drift — fine at the start, badly off an hour in. The subtitle file was timed against a version at a different frame rate. The classic case is film shot at 24 fps and sped up to 25 fps for PAL broadcast, which makes the video about 4% shorter. Subtitles timed for one run steadily out of step against the other. Most players have a speed or frame-rate adjustment for exactly this.
Constant offset — consistently a second or two out, but never worsening. The file is timed for a version with a different opening: a studio logo, a different length of titles. Every decent player lets you nudge subtitle delay with a keyboard shortcut, and one adjustment fixes the whole file.
Correct for a while then suddenly wrong — the file matches a version with different advert breaks or an extra scene. Harder to fix; usually the answer is a different file.
Formats, briefly
- SRT — plain text with timestamps. Universally supported, almost no formatting. The safe default.
- ASS/SSA — supports positioning, fonts, colour and effects. Common in fansubbed animation. Less widely supported on TVs.
- VTT — the web standard, used by browser video players.
- PGS — image-based subtitles from Blu-ray. Pictures rather than text, so they can't be restyled or searched, and they're much larger.
Text formats are preferable where you have the choice: smaller, restylable, and your player can adjust size for readability.
Burned-in versus selectable
Burned-in subtitles are part of the picture. They always display and always work, and you can never turn them off or change their appearance.
Selectable subtitles are a separate track the player renders. Toggleable, restylable, and dependent on the player supporting the format.
Burned-in is also why some streams force a transcode when you enable subtitles — the server has to re-render the video with the text in it, which on a home media server can turn a light job into a heavy one.
Readability settings worth changing
Most people leave defaults that are smaller and harder to read than necessary. In your player or platform's accessibility settings you can usually adjust size, choose a font with clearer letterforms, and — the one that helps most — add a solid or semi-opaque background box behind the text.
Text with an outline sitting directly over a bright scene is the common failure. A background box fixes it completely and costs nothing.