Why a plan helps
Subtitling looks like a small task at the end of production: add some text to a finished video. In practice it has several stages that depend on each other, and problems found late are usually the most expensive to fix. A short plan written before work starts tends to save more time than it costs.
The stages below are a general outline. Smaller jobs may combine or skip some of them, and larger ones may add more.
Step 1: Decide what the subtitles are for
Start with the purpose and the audience. Are the subtitles meant to help viewers who are deaf or hard of hearing, to translate speech into other languages, to support viewing with the sound off, or a mix of these? The answer shapes almost every later choice. Captions that include non-speech information serve a different need than plain translated dialogue, a difference explained in the discussion of captions versus subtitles.
Also note where the video will be shown. A short clip for social video, a course lecture and a broadcast program can have quite different expectations for style, format and checking.
Step 2: Define the scope
Write down the basics:
- how many videos, and their combined running time
- the source language and each target language
- whether the subtitles are same-language captions, translations, or both
- the delivery format and how they will be attached to the video
- who will approve the result, and by when
Requirements in some places, especially for public bodies and broadcasters, may apply to the work. These vary by country and change over time, so check with the relevant authority or a qualified professional rather than assuming.
Step 3: Prepare the source
The quality of the input limits the quality of the output. Gather the final, locked picture, a script if one exists, and any lists of names, terms and brand spellings. Good source preparation reduces questions later. Clear recorded sound also matters, particularly if a first draft will come from speech recognition, because poor audio quality leads to more errors to fix.
If the video might still change, say so now. Edits after subtitling mean rework, which is its own topic when a video is re-edited.
Step 4: Choose how the work will be done
Decide who does what. Some teams handle everything themselves, some buy in the whole job, and many split it. Automated tools can produce a rough first transcript or draft translation quickly, but their output is a starting point that people need to check. Whether to keep the work inside the team or hand it out is a decision with real trade-offs, covered separately in the outsourcing decision article.
Step 5: Transcribe or script, then segment
Subtitling begins from the words spoken. That may be an existing script, a transcript made by a person, or an automatic draft that is then corrected. The text is then divided into cues, each with a start time, an end time and one or two lines of text. Dividing the speech sensibly, and deciding where each piece appears, is often called the spotting process.
Writing the text means fitting speech into limited space and time. Following a set of style guidelines, covering line length, reading speed and punctuation, keeps the result consistent, especially when several people contribute.
Step 6: Translate, if needed
Translation adds its own steps. Translators work from the script or the audio, and the choice affects accuracy, as discussed under source material. Text often grows or shrinks between languages, so text expansion can force rewrites to fit. A shared glossary and, where available, a translation memory help keep terms and repeated phrases consistent across a project.
Step 7: Time and sync
Each cue needs to appear and disappear at sensible moments, which is the subject of subtitle timing. Cues are usually checked against dialogue, scene changes and the minimum and maximum time on screen. Frame rate mismatches can cause sync drift if they are not caught, so confirm the video’s frame rate before timing begins.
Step 8: Review and quality control
Allow real time for checking. A typical pass covers spelling and grammar, names, timing, line breaks and formatting, ideally using a written quality checklist. For important material, such as medical or legal content, have a qualified human reviewer read the final text; automated checks cannot catch every error. A separate person from the writer usually finds more problems.
Step 9: Deliver and publish
Export the files in the agreed format, check character encoding if non-English text is involved, and test the subtitles in the real player or platform. Confirm that each language is labeled correctly and that the subtitles can be turned on and off as intended. If they are burned into the picture instead, check every frame range once more, since they cannot be changed afterward.
Step 10: Keep records
Store the approved files, the glossary and the notes on decisions. If the video is reused, updated or translated into another language later, these records save a great deal of effort.
Build in time and buffer
Estimates are rarely exact. The time a job needs depends on length, audio, number of languages, review rounds and who is available, as described under turnaround time. When in doubt, add a margin for questions, corrections and approval, and agree the schedule with everyone involved before work starts.