First, establish which materials are available
We ask for the final video edit, source language, target languages and publication method. We check whether a script, continuous text without timecodes or source SRT / WebVTT subtitles are available. Each starting point needs slightly different preparation.
If the client supplies text without timecodes, we can segment it into subtitles and synchronize it with the recording. Where subtitles already exist, we verify them against the video. Before quoting, we also agree linguistic review scope, approval planning and final files.
AI / STT transcription and text verification
If no source text is available, we prepare automatic AI transcription using speech-to-text (STT). With clear speech and good audio, accuracy is usually high, reducing manual transcription and preparation time. Accents, noise, overlapping speakers and specialist names can increase correction needs.
Generated text always needs verification; raw STT output is not treated as translation-ready content. We offer in-house linguistic verification in Polish and English. Other languages are referred to the client’s designated reviewer or SME.
For review, we can supply text or a video with temporary visible subtitles in Frame.io. The latter brings visuals, voice and text together. Product names, numbers and specialist terms are checked against the project glossary.
Timing, segmentation and existing subtitle checks
We align segment in and out times with speech and editing. We check gaps, overlaps, display duration and reading speed. We agree a validation profile including characters per line (CPL), line count and reading speed (CPS / WPM).
We refine segmentation by joining fragments into logical sentences, splitting longer passages and considering punctuation and natural phrase boundaries. Punctuation alone does not automatically determine a break; meaning, visuals and reading time also matter.
We can perform the same stage on client-supplied subtitles. There is no need to transcribe again if the existing text is suitable; we can handle timing checks, segmentation and preparation for further localization.
Handover of prepared files for translation
We deliver corrected source subtitles to the client’s language team. We agree the format: SRT, VTT, CaptionHub, CAT tool import or Excel. We include instructions for names, untranslated interface elements and readability requirements.
Translation and approval of target content remain with the client. If review takes place in Excel, we prepare suitable columns and IDs so approved changes can be returned to the subtitle file. The agreed structure, sequence and version labels should be retained.
Post-translation checks and SME review in Frame.io
After receiving the translation, we check file structure, text segmentation and synchronization. Target text may be longer than the source, so retaining identical timecodes does not guarantee readability. Necessary shortening or content changes are submitted for linguistic approval.
We prepare a video with temporary subtitles and share it in Frame.io. The SME watches the visuals, listens to the recording and reads the translation, commenting on specific moments. This helps reveal context missing from a text file alone.
For example, the Polish translation uses the masculine “wybrałem,” but the person speaking on screen is a woman. The reviewer requests the feminine “wybrałam” and confirms that Dashboard should remain in English because the application has no Polish version. We implement approved feedback and recheck the subtitles.
SRT / VTT or burned-in subtitles
The client can receive a separate subtitle file or video with burned-in subtitles. SRT and WebVTT suit players that support separate text tracks. Burned-in subtitles become part of the image and remain visible regardless of player settings.
For burn-in, we prepare fonts, size, contrast, outline, background and margins. We can reposition selected segments when they collide with on-screen text (OST), speaker captions or other important visuals. Style and positioning are approved on a sample.
We work in tools including CaptionHub and Subtitle Edit; burn-in can use an agreed ASS / SSA workflow or Adobe Premiere Pro. Before delivery, we check the final file or render, language completeness, encoding and version labels. Changes to the video edit require timing to be reassessed.
What should you prepare for analysis and a quote?
- Final video or link, duration and edit version
- Script, continuous text or existing source subtitles
- Languages, terminology glossary and designated reviewer
- SRT / VTT or burn-in, readability guidelines and deadline
Send the available materials and requirements. We will identify missing items and agree scope, cost and deadline before starting.
Discuss your project ↗Frequently asked questions
Can we start with plain text without timecodes?
Yes. Using the text and recording, we prepare segments, timing and an SRT or VTT file. We verify the text against what is actually said.
Do you check subtitles prepared by the client?
Yes. We can handle synchronization, segmentation and readability checks. In-house linguistic verification covers Polish and English; other languages are checked by the designated reviewer or SME.
Do subtitles always have to be burned in after translation?
No. Burn-in is one delivery option. We can provide separate SRT / VTT files; a working video with visible subtitles can be used solely for review in Frame.io.
Tool documentation
The process above describes how we work at KeboDesign. The following vendor resources provide additional information about the tools’ features.