Subtitle translation workflow

Translating subtitles is not the same as translating a script. The translator must work within the constraints of timing, line length and reading speed, while preserving the meaning and tone of the original. This is the practical workflow professionals use.

Input formats

Translators usually work with one of two starting points:

• Source subtitle file (SRT or VTT). You have the original dialogue with timings. You translate cue by cue, keeping the same timings.
• Source transcript or script. You have the dialogue as plain text, without timings. You must produce a subtitle file from scratch, including timing.

The first case is far more common and far faster. If you only have a transcript, the timing work is much larger.

The efficient workflow

1. Prepare the source file. Load it into your translation tool of choice (a subtitle editor, or a spreadsheet if translating line by line).
2. Translate cue by cue. Preserve timings from the source. Adjust text length as needed to fit reading speed.
3. Check character limits. The translated language may be longer or shorter than the source. Use the Line Length Checker to find cues that exceed the target.
4. Validate. Run the file through the SRT Validator to catch timing issues introduced during translation.
5. Deliver. Most streaming platforms accept SRT or WebVTT.

Translation-specific challenges

Language length differences

Different languages have different character-to-meaning ratios. German and French tend to be longer than English. Chinese and Japanese are far denser — a single CJK character can carry what takes several Latin characters to say.

The implication for translation:

• Translating from English to German — expect 20–35% longer text. Many cues will need to be split or compressed.
• Translating from English to Japanese/Chinese — expect significantly shorter text. Cue length becomes less of a concern, but speaker labelling and formatting rules differ.
• Translating from English to Arabic/Hebrew — right-to-left rendering matters. Line length and punctuation interact with direction.

Cultural references and wordplay

Some phrases cannot be translated literally. Idioms, puns and cultural references must be adapted for the target audience. The translator's job is to preserve the meaning and effect, not the exact words.

Reading speed across languages

Reading speed targets (CPS) are language-dependent. English readers average around 15 CPS for subtitles. Japanese and Chinese readers can read denser text — the effective CPS target may be lower in character count but higher in information density. Follow the target language's own guidelines.

Multi-speaker content

If the original marks speakers (e.g. - MARY:), preserve the labels. If the original uses voice tags (<v Mary>), they must be preserved through translation. Never remove speaker information.

Sound effects and SDH

If translating SDH subtitles, the sound effect descriptions in brackets must also be translated. Localising sound effects is subtle — some sounds are language-specific (a knock in one language is not universal), and sometimes the right choice is to adapt rather than translate literally.

Bilingual subtitle deliverables

Some content needs both the source language and translation visible simultaneously, usually stacked in the same cue:

1
00:00:01,000 --> 00:00:04,000
Hello and welcome.
Bonjour et bienvenue.

To create bilingual files:

1. Translate the source to produce a second subtitle file with the same timings.
2. Merge the two files line by line. The Subtitle Merger does chronological merge, not line-by-line stacking — for line-by-line stacking you will need to combine manually or with a script.
3. Validate the merged file with SRT Validator.

Quality checks

• Completeness. Every cue has a translation. No cues left in the source language by accident.
• Consistency. Proper nouns, character names and terminology are consistent across cues.
• Length. Translated cues respect character limits.
• Reading speed. Translated cues do not exceed CPS targets.
• Timing. Timings are preserved from the source.
• Formatting. Speaker labels, brackets and italics preserved.

Tools for subtitle translation

• SRT to Plain Text — extract dialogue for a translation memory.
• Search & Replace — batch-correct consistent terms.
• Line Length Checker — flag overlong lines.
• SRT Validator — final check.
• SRT to VTT — prepare for web delivery.

Further reading

• Accessibility guide
• Line length guidelines
• Reading speed guide