Runs on our Mac
Whisper (faster-whisper) and NLLB run locally. Your footage is never sent to a cloud service.
Drop a video or paste a TikTok, Instagram or YouTube link. Whisper runs on our own Mac to write the captions; edit them, translate them, and burn them in TikTok style.
Files deleted after an hour
Drop a file, or paste a public TikTok, Instagram or YouTube link. Pick Fast for quick drafts or Accurate for tricky audio.
The spoken language is detected for you. Fix any word in the editor and press Save; the preview updates straight away.
Download SRT, VTT or a plain transcript, translate it, or make an MP4 with big word-by-word captions.
Whisper (faster-whisper) and NLLB run locally. Your footage is never sent to a cloud service.
Three model sizes: base, small and medium.
Big, bold, centred words with the spoken word highlighted, in three presets.
Translate captions to English, Malay, Chinese or Thai.
Every line is editable in the browser, in every language.
Uploads and results are deleted an hour after your last step.
Video: MP4, MOV, M4V, WebM, MKV. Audio: MP3, M4A, WAV, AAC, OGG, FLAC. Up to 500 MB and 15 minutes. Audio files get captions and transcripts but no burned-in video.
On Fast, a one-minute clip is done in a few seconds; Accurate is several times slower. The first job with a model also loads it, which adds a little time. Only one or two jobs run at once, so you may wait in a short queue.
They come from NLLB-200 (600M), an open model running on our Mac. English and Malay are usually solid; Chinese and Thai are readable but need a native speaker's check before anything is published. We do not send text to Google or any other translation API.
Chinese is highlighted character by character. Thai is written without spaces between words, so the highlight moves by phrase.