Multilingual subtitles: one video, up to 20 languages
Drop a video, an audio file or an SRT / VTT file and get a subtitle pack in every language you pick — same timing in each one, line length and reading speed checked per language, side-by-side editing, a ZIP of SRT or VTT files, and YouTube caption tracks through the free Scheduler. Transcription runs on your device; translation runs on your device too when your browser can.
Free · No sign-upHow to make subtitles in several languages at once
- 1. Drop a video or audio file and press Transcribe — Whisper writes the source subtitles on your device (the model downloads once, 41–250 MB, then it's cached). Already have subtitles? Drop the SRT or VTT file, or paste it.
- 2. Pick up to 20 languages (or a preset: EU pack, top 10 world languages, Spanish + Portuguese) and press Translate. Each subtitle is translated on its own, so the timing stays exactly as it was.
- 3. Review side by side. Lines that are too long for two lines or too fast to read are flagged with the guide for that language. Edit any line; with a video you can preview the captions burned into the picture.
- 4. Download a ZIP with one SRT or VTT file per language (named like
video.es.srt, which YouTube and most players recognise), or Send to Scheduler to add them as caption tracks to your YouTube video.
What makes a subtitle easy to read in every language
Good subtitles are short and stay on screen long enough to read. The pack checks every line against a guide for its language: about 42 characters a line and 17 characters a second for most languages, 16 characters a line for Chinese and Korean and 13 for Japanese, which pack more meaning into each character. Thai is written without spaces, so lines are broken between words found by your browser's word segmenter, and Chinese and Japanese lines never start with a comma, a full stop or a small kana.
Arabic, Hebrew, Persian and Urdu are written right to left. The review grid shows them right to left, and the SRT files get invisible direction marks around each line so players without full bidirectional text support still put the punctuation on the correct side. VTT files and YouTube caption tracks don't need them.
Want viewers to hear it in English instead of reading it? The AI Video Dubbing tool turns the same subtitles into an English voice-over track with a stock AI voice, timed to the original.
Machine translation is a strong first draft, not a finished translation. Names, jokes and technical terms deserve a second look — track names say “auto-translated” by default so viewers know. For one document or a single subtitle file in one language, the AI Translator is quicker; to create subtitles in the video's own language, try the Text-Based Video Editor or the Video Trimmer & Subtitles.
Frequently asked questions
Does translating change the subtitle timing?
No. Every language gets exactly the same cues with the same start and end times as the source; only the text is translated, one cue at a time. The only optional change is “Give fast subtitles more time”, which extends the end of a subtitle into the pause after it — start times never move.
Where does the translation happen?
In Chrome with built-in AI, Chrome's translator runs on your computer and nothing is sent anywhere. Otherwise an on-device model you downloaded, your own free AI key (sent straight from your browser to that provider) or GrabCast's free AI with a daily cap is used. A chip next to each language shows which engine translated it.
How do the line length and reading speed checks work?
Each language has a guide from common subtitling practice: for most languages about 42 characters a line and 17 characters a second, 16 characters a line for Chinese and Korean, 13 for Japanese. Chinese and Japanese are broken by character and never start a line with a comma or full stop, Thai is broken between words, and Arabic, Hebrew, Persian and Urdu are handled right to left. Lines outside the guide get a warning you can fix in the grid.
Can GrabCast add the subtitles to my YouTube video?
Yes, for your own channel. Press Send to Scheduler, then attach the pack to a YouTube post. After YouTube has processed the video, GrabCast uploads one caption track per language, optionally as drafts. It needs the YouTube permission that also covers comments, and each track uses part of GrabCast's daily YouTube quota, so the Scheduler shows how many tracks are left today.
Is my video uploaded?
No. The speech is transcribed by Whisper inside your browser; the only downloads are the speech model (41–250 MB, once, then cached). Only the subtitle text goes to a translation service when Chrome's on-device translator isn't available, and nothing is uploaded to GrabCast unless you press Send to Scheduler.
Is there a watermark or a limit?
No watermark and no length limit. You can pick up to 20 languages per run. GrabCast's free AI has a daily cap; Chrome's on-device translator and your own free key have none on our side.