Subtitle translation: from transcript to bilingual output

Most subtitle quality is decided before translation starts: how accurate the transcript is, and how consistent the terminology is.

The chain

  1. Transcript. Start from a proofread transcript with timing (see the subtitle generation guide).
  2. Context grouping. Translate a few neighbouring lines together — line-by-line translation breaks tone and pronouns.
  3. Terminology. Fix names, brands and technical terms once, then apply them across the whole file.
  4. Translate. Produce the target language and keep line length readable.
  5. Lay out bilingual output. Original on top, translation below; check size so the picture is not obscured.

Common pitfalls

PitfallSymptomFix
Line-by-line translationBroken tone, ambiguous pronounsTranslate in context groups
Terminology driftThe same term translated differentlyBuild a term list and enforce it
Timing shiftText out of sync with speechKeep the original timing; replace only the text
Overlong linesViewers cannot read fast enoughCap characters per line and wrap sensibly

Bilingual layout

  • Default to original on top, translation below.
  • Slightly smaller type than body text so it does not dominate the frame.
  • On busy backgrounds, a subtle outline improves legibility.

Frequently asked questions

What causes the worst subtitle translations?

An inaccurate transcript and inconsistent terminology. If the source text is wrong, the translation will be too.

Why not translate one line at a time?

Line-by-line translation loses context, so tone and pronouns drift. Translating a few neighbouring lines together reads far better.

How should bilingual subtitles be arranged?

Original on top, translation below, slightly smaller, with a sensible character limit per line.

More topics on the guides hub, or see what Lumasce does.