Running time is only part of the story
It is tempting to assume that a ten-minute video takes a fixed multiple of ten minutes to subtitle. In practice the same length can take very different amounts of time. Content, quality of the source, the number of languages and the amount of checking all matter, often more than duration. No single ratio applies to every job, so treat any quoted figures as rough guides and expect them to vary.
This article lists the main influences. They are grouped by where they arise.
The material itself
Density of speech. A video with continuous, fast talking produces many cues and demands more condensing. A slow, sparse video with long pauses is quicker. Reading speed limits mean fast dialogue needs more editing decisions.
Number of speakers. Overlapping voices, interruptions and frequent speaker changes take longer to sort out, label and time. Showing speaker changes clearly is extra work.
Audio quality. Clear, close-recorded speech is quick to transcribe. Echo, background noise, music beneath speech or distant microphones slow people down, and also reduce the accuracy of automatic drafts. Replaying difficult passages again and again adds up.
Accents and speaking style. Strong regional accents, mumbling or unusual vocabulary can need repeat listening. Accent recognition problems affect software as well as people.
Specialist content. Technical, medical or legal talks need research and careful term checking. A glossary helps; without one, each unfamiliar word can mean a pause to look it up.
Songs, wordplay and cultural references. Material that cannot be translated literally takes thought, discussion and sometimes a query to the client.
The source preparation
A finished picture, a script, a name list and notes make everything faster. The opposite, a rough cut that may change or a missing file, causes delays. Waiting for answers to questions can stretch a schedule more than the work itself, so a reachable contact makes a real difference. Good source preparation is one of the cheapest ways to save time.
The type of work
Same-language captions versus translation. Translation adds writing, and also checking by someone who reads the target language. Each extra language adds its own cycle of work and review.
Number of languages. Languages can often be worked in parallel, but only if enough qualified people are available. Rare language pairs may have fewer available specialists, and scheduling can take longer. A note of caution applies to translating through an intermediate language, which introduces its own delay and risk of error.
Captions with sound information. Adding descriptions of sound effects and music needs extra viewing and decisions about what to include.
Language characteristics. Some languages need more space for the same meaning, so text expansion forces rewriting. Others need special handling of line breaks, scripts or direction of text, which adds checks.
The method and the tools
Automatic first drafts. Speech recognition and machine translation can produce a quick starting point. Whether this saves time depends on the quality of the draft: if it is poor, correcting it may take as long as starting fresh. Machine output always needs human checking, and it should not be treated as finished work. How efficiently people can edit automatic captions also depends on their tools and habits.
Tool familiarity. A skilled person using a tool they know well works faster. Switching tools or formats mid-project costs time.
Translation memory and glossaries. Reusing earlier approved wording speeds up repeated content and keeps terms consistent.
Review and approval
This is the stage most often underestimated. Allow time for:
- a proofreading pass for spelling, grammar and timing
- a check by a fluent reviewer in each language
- a quality checklist run before delivery
- feedback from the client or subject experts, and the corrections that follow
Every round of feedback adds time, and a long wait for approval can cost more than the work. Important material, such as medical or legal content, should have a qualified human reviewer, and this should be planned into the schedule from the start. Fewer, clearer review rounds are better than many informal ones.
Delivery and technical requirements
Exporting, checking that files play correctly and fixing encoding or format problems takes time, especially with several formats or platforms. If subtitles are burned into the picture, rendering a new video and checking it adds more. Meeting a strict delivery specification can mean extra checks that a casual job would skip.
Changes during the job
A re-edit, a changed script or a new term list partway through usually causes rework. The earlier the picture is locked, the lower this risk.
Workload and availability
The people doing the work have other jobs too. Busy periods, holidays and time zones affect when work can start and finish. A request that includes an urgent deadline may be possible, but rushing can leave less time for checking, so quality can suffer. If you need a fast turnaround, discuss what review will still happen.
Setting a realistic schedule
A sensible approach is to ask for an estimate that covers each stage separately, share any known difficulties up front, and add a margin for questions and corrections. Where possible, try a short sample first, since it shows how demanding the material really is. Outcomes are uncertain, so agree what happens if the work runs late.