How to add subtitles in another language to a video on Windows
You recorded a walkthrough in English and the people who need it read Spanish, or Turkish, or Japanese. Rewriting it is out of the question. Captions in their language are enough - if you can get them without typing every line and lining it up with the video by hand.
AgentEyes, our screen and audio recorder for Windows, does it in two steps: translate the transcript, then burn the captions onto a new copy of the video.
Step zero: a transcript
Translation starts from a transcript, so the recording needs one first. A recording you made in AgentEyes is transcribed after you stop it, as long as you are signed in; if that step did not run, build its walkthrough. A video from somewhere else - a Teams meeting recording, a capture from another tool, a clip someone sent you - comes in with import:
agenteyes import "C:\Users\me\Videos\Weekly sync.mp4"- It copies, it does not move. Your original file stays exactly where it was. The copy goes into a new folder in your AgentEyes library.
- Most containers work. MP4, MKV, MOV, AVI, WebM, WMV and the other common ones. The file needs a video stream; an audio-only file is turned away with a clear message.
- You get the transcript three ways. Timed segments in
transcript.json, plain text intranscript.txt, and captions intranscript.en.vtt.
Import prints the new recording's id when it is done. Keep it; the next two commands take it.
Step one: translate, with the timing kept
agenteyes translate <id> --to esEach caption keeps its original start and end, so the translated captions line up with the video cue for cue. Use a short language code such as es, tr, fr, de or ja, or a regional one like pt-BR. The result is another file beside the first, transcript.es.vtt, and a recording can carry as many languages as you translate it into.
Those are standard WebVTT files, the caption format browsers understand, so you can stop here if a caption file is all you need. Translate is a command-line and API step; there is no button for it in the app.
Step two: burn the captions in
Burning writes the captions onto the picture of a brand-new video, for places that will not take a separate caption file. Two ways to do it:
- In the app. Right-click the recording in the library and choose Export with subtitles.... It offers only the languages that recording carries, burns the one you pick, and opens the folder on the new file.
- From the command line.
agenteyes subtitle <id> --lang es
Either way you get recording.es.subtitled.mp4 in the recording folder. Your original video is never touched. If you ask for a language the recording does not have yet, it stops and tells you to translate that language first, rather than producing a video with no captions.
What is sent where
Worth knowing before you point it at a meeting recording:
- Transcribing uses DevThrottle's hosted AI. Import extracts the audio and sends that, through your signed-in account. The video file itself is not part of the request.
- Translating sends the transcript text, again through your signed-in account.
- Burning runs on your own machine with the bundled ffmpeg. Nothing is sent anywhere for that step.
The whole flow, including the same two steps over AgentEyes' local Control API, is on the Multilingual subtitles page. For bringing outside videos in, see Import a video.