Do you preserve timecodes exactly?

Yes, byte-for-byte. Timecodes are never sent to translation.

Can I set a target CPS limit?

Yes. Defaults to 17 (Netflix standard). Set higher for broadcast or lower for accessibility.

What about line length per row?

Default 42 characters. Configurable at upload.

Speaker tags and italic?

Both preserved. [MAN], [WOMAN], dashes, music notes, and / tags.

Do you handle VTT (WebVTT) too?

Not yet — only SRT. WebVTT support on the roadmap.

Can I send audio and get SRT back?

Yes — upload audio and select SRT as the output format. The audio transcription pipeline writes timestamped SRT.

Subtitles · AI translation

Translate SRT subtitles. Timecodes and line breaks preserved.

Drop your .srt and get it back translated with timecodes intact, line breaks adjusted for the target language, and character-per-second limits respected.

See how it works

What is SRT Subtitles?

SRT (SubRip) is the most widely used subtitle format. Plain text with numbered cues, each cue containing a timecode range and 1–2 lines of text. Translation of SRT files requires more than swapping text — line length, reading speed (characters per second), and line break placement all need adjustment for the target language.

Why SRT Subtitles is tricky for AI translation

Timecodes must remain byte-for-byte exact. A single character drift breaks sync with the video.

Character-per-second (CPS) reading-speed limits differ by platform: Netflix 17 CPS, YouTube no hard limit, broadcast 20+ CPS. Translations must respect target CPS or viewers can't read in time.

Line length per row matters: 42 characters max per line on Netflix, 32 on some broadcast standards.

Text expansion (Spanish +25%, German +30%) means the translation may exceed CPS or line length — needs paraphrasing.

Speaker tags ([MAN], [WOMAN], dashes for dialogue) must be preserved across translation.

Italic and HTML tags (, ) in some SRT dialects need to stay intact.

How Fily handles SRT Subtitles

Timecodes pass-through: timestamps are extracted as-is and never sent to translation.

CPS-aware translation: target CPS set at upload; if a translation exceeds the limit, the model is prompted to paraphrase tighter.

Line break tuning: line breaks placed at natural pauses (clause boundaries, breath points), not mid-phrase.

Speaker tags preserved: [MAN], [WOMAN], dashes, and music markers (♪) all pass through.

Inline tags preserved: , , survive translation.

Output validated: every cue checked for valid timecode order and non-empty text before delivery.

Pipeline: srt_qa@1.0.0

The SRT Subtitles workflow with Fily

Upload

Drop your .srt (single or batch ZIP). Optional: glossary, TM, style guide.

Process

Fily runs the SRT Subtitles pipeline + 12 QA steps. Typical job: 10–20 minutes.

Download

Same format, ready to deliver. QA report HTML attached.

Common upload: a 45-minute documentary's SRT file in English, needed in Spanish for LATAM distribution. Fily delivers the SRT with timecodes intact, line breaks adjusted for Spanish's longer phrasing, and CPS staying under Netflix's 17.