Can you use an SRT file as a dubbing script?

Use subtitles as a reference, then establish what is missing

An SRT file can be a useful starting reference for a dubbing team, but converting it to text does not make it a finished dubbing script. First compare its wording with the exact recording, identify the speakers manually and separate subtitle display times from speech boundaries. Then ask the recipient which document and format they actually need. Preserve the original SRT throughout.

This guide is for recordings and text you own or have permission to adapt and share. The result is a reviewable reference package for a translator, adaptor or studio. Voice performance, recording, synchronization and final delivery approval are separate work. If the recipient requires a specific production template, obtain it before reorganizing hundreds of cues.

Name the document and the audio it describes

A source-dialogue reference records the words in the original audio and helps the next person locate them. An adapted recording script proposes words to be performed in the target language. An as-recorded script describes what the approved final dub actually contains. These documents can differ even when they refer to the same picture.

Netflix provides a concrete example in its As Recorded Dubbing Script Scope of Work: its source Dialogue List and target-language as-recorded script have different roles, and the latter must follow the final audio and be delivered in TTAL. These are Netflix requirements, not a universal format or a claim that any SRT export will be accepted. Follow your own recipient’s current specification.

Put the stage, language, video revision and audio version at the top of your reference. A translated subtitle track should not be labeled a transcript of a dub that has not been recorded. Likewise, an earlier rehearsal recording should not silently stand in for the approved final mix.

Check completeness before changing the layout

Listen to the matching original audio while reading the source subtitles. Captions may contain edited phrasing, fragments split across cues, speaker labels or sound descriptions. A translated track may paraphrase the original. Establish which kind of text you received before treating it as a word-for-word account.

W3C’s transcription guidance calls for correction of automatic drafts and accurate representation of the audio. In this workflow, apply that review to every passage, including repetitions, negatives, names and short interjections. Keep uncertainty visible in a separate note. An unclear word does not become reliable when placed in a spreadsheet.

Classify non-dialogue text deliberately. A caption such as [door closes] is not a spoken line for an actor. On-screen lettering, reactions and off-screen speech may still matter to the production, so ask where the recipient wants them recorded. Preserve these elements in the original caption file; building a dialogue reference is not a reason to strip information from the accessibility deliverable.

Example: one subtitle cue can hide two recording lines

The following original example is fictional. Imagine an approved scene in which Mara says “No, leave the green lamp on.” from 12.200 to 13.800 seconds, followed by Ivo saying “All right.” from 14.100 to 14.700. The subtitle editor has combined the exchange into one cue and shortened the first line. Those invented audio observations are supplied for the example; they were not measured in a real recording.

The SRT identifies a five-second display interval. It does not identify the two characters, recover the omitted “No,” or tell you exactly when either voice begins. Copying its interval into two recording rows would give both characters the same boundaries and conceal the pause. A line-leading dash suggests a turn but does not name the speaker.

Received subtitle excerpt

42
00:00:12,000 --> 00:00:17,000
- Leave the green lamp on.
- All right.

Create a separate, traceable dialogue reference

In a separate document, retain the source cue identifier and its display interval, then add the words and boundaries verified against the recording. Give each manually identified speaker a separate reference entry. If a sentence spans several cues, list all contributing cue IDs. If two voices overlap, preserve that relationship rather than forcing them into a false sequence.

The example below keeps cue 42 as the origin of both entries. Its original subtitle text and timing remain untouched in the saved SRT. The reference restores “No,” because the fictional audio establishes it, and records separate speech boundaries. These rows are a suggested review aid, not a studio delivery format or an automatic Polyglot export.

Use elapsed media time unless the recipient explicitly asks for another timebase. If they require frame-based timecode, agree the frame rate, starting timecode and conversion convention separately. SRT millisecond values are not frame numbers. Do not move the subtitle file to imitate recording timings merely to make the two documents look alike.

Manual source reference
Video: lamp-scene-v3
Audio: original English v3
Origin: SRT cue 42
Display: 12.000–17.000 seconds

Entry A | Character: Mara
Speech: 12.200–13.800 seconds
No, leave the green lamp on.

Entry B | Character: Ivo
Speech: 14.100–14.700 seconds
All right.

Target adaptation: pending
Recording approval: pending

Keep translation and performance decisions separate

A reviewed subtitle translation can help explain meaning, but it should not automatically become the actor’s recording text. Read the proposed target dialogue aloud against the picture, with a qualified language reviewer and the production team. Check intent, pronunciation, character voice and the available performance space. Preserve the source wording beside the proposal so an intentional adaptation remains distinguishable from a transcription correction.

Netflix’s English film-and-series dubbing guidelines discuss the balance between intended meaning and lip synchronization. That illustrates why preserving an SRT cue’s duration is not a test of dubbing readiness. Two sentences can fit the same number of seconds yet have different rhythm, emphasis or visible mouth movements.

After recording, compare the approved target audio with the adapted script. Performers may have used different words or pauses. Update a separately identified as-recorded document from that audio; do not overwrite the original-language reference. Mark unrecorded changes as proposals until the relevant reviewer approves them.

Use Polyglot for the subtitle work it supports

Polyglot can import an existing SRT, let you review and edit its text and timing, translate subtitles and export SRT, WebVTT or a text transcript locally. Keep the reviewed timed file beside any text export. A text transcript can help someone read the material, but it does not supply a complete studio dialogue list with verified character identities and speech boundaries.

This workflow does not promise automatic dubbing, generated voices, lip-sync adaptation, speaker identification, TTAL export or a Netflix integration. Prepare the manual reference and any recipient-specific production document outside the subtitle export. Preserve completed translations and unsaved edits; a new recording-script revision is a separate artifact, not permission to replace approved subtitle work.

If the final dub also needs captions, review a track against that exact target-language audio. Translating subtitles for the original soundtrack and captioning what was actually performed in the dub are different acceptance checks. Retain both audio-version labels so the next person can choose the intended track.

Hand over a reference with an explicit review status

Send only the permitted media and text versions, the untouched source SRT, the reviewed manual reference and a short list of unresolved passages. Name the reviewer responsible for character identity, wording, timing and target adaptation. Keep private production notes separate from material meant for viewers.

Before calling the package ready, have the recipient confirm that they can find every passage in the agreed media and that the document fits their current process. A successfully opened text file establishes readability, not studio acceptance. If only the source review is complete, say so. Leave adaptation, recording and final-script approval pending until those stages have actually happened.

Handoff status
Media and audio revision: ____
Source language: ____________
Source SRT retained: ________
All dialogue reviewed: _____
Characters verified: _______
Speech timing reviewed: ____
Recipient format agreed: ___
Unresolved passages: _______
Adaptation approval: _______
Final recording checked: ___

Keep exploring.