How to translate video subtitles without breaking the timing
Translation is the easy half. The half that ruins subtitle files is timing — and it breaks in a very specific way, so it is worth knowing before you start.
The failure everyone hits
You take a transcript, paste it into a translator, get English back, then put it in an SRT. Two things go wrong at once: the sentences come back merged or split differently than the subtitles need, and if the timestamps were in the text you pasted, they got translated or reordered too. The result is English subtitles that drift, jump, or sit on screen while nobody is talking.
The three steps that avoid it
- Transcribe the audio, keep the timing. Word-level timing from the actual speech. This is the only reliable source for "when did they say this".
- Translate one line at a time. Every line keeps its original start and end. Do not translate the whole file as one block, and never let the tool re-time the output.
- Render or export. Either burn the subtitles into the video, or export an SRT and upload it alongside on YouTube.
Try it — free under 60 seconds, no account
Two details that matter in practice
- Line length. English runs 20-40% longer than Korean, Chinese or Japanese. If you keep the same character count per line, the text will overflow. Split into readable two-line blocks instead of shrinking the font.
- Read it before publishing. Automatic transcription guesses at names, brands and accents. A two-minute read catches almost all of it.
Doing it by hand
Whisper handles step one locally if you have a machine for it. Step two is usually where people give up — a 20-minute video is a few hundred lines. Step three is ffmpeg or the YouTube subtitle uploader.
This site does the three steps in one pass: upload, wait, download the SRT or the burned-in video. Free under 60 seconds, so you can check the quality on your own footage first.