Skip to content
Start a project See the work

Bilingual captions with every name spelled right.

They wanted English and Chinese captions that use the real script, so we built a tool that times subtitles with speech recognition and burns the exact wording into talking-head video.

Captionsinterview-final.mp4Export videoWe start from the script, not a guess.我们从脚本出发,而不是靠猜。00:42ScriptEN · 中文01Two languages,one edit.两种语言,一次剪辑。02We start from thescript, not a guess.从脚本出发,不靠猜。03Every name isspelled right.每个名字都拼写正确。04Timing is checkedline by line.逐行核对时间。05Then one exportper language.每种语言各导出一次。06Ready for everyplatform.适用于所有平台。00:3000:3400:3800:4200:4600:5000:54ENZHWe start fromthe script,not a guess.Every namespelled right,in bothlanguages.我们从脚本出发而不是靠猜每个名字都拼写正确两种语言都是

Screen redrawn with sample data. The client is not named.

Customer
Video creators filming in two languages
The result
Captions match the script word for word, in both languages.
Runs on
Desktop
Built with
Python, ffmpeg
What they wanted

Auto-captions kept misspelling names and products, and fixing them by hand took longer than the edit. They already had the script and wanted captions that used it, in English and Chinese.

How we approached it

We use speech recognition only for timing, then lay the real script and its translation onto it.

What we delivered
  • Upload a video, a script and its translation
  • Pick the paragraphs to caption
  • Alignment that follows the speaker
  • Preview, then download the finished video

Have something that should work better?

Tell us what is getting in the way. A person reads it and replies within two business days.