Generate subtitles from your video transcript automatically
A video without subtitles loses a large share of its audience: muted phones, viewers without headphones, people with hearing impairments. Typing and timing subtitles by hand is still one of the least popular steps in editing.
Why is manual subtitling such a slog?
Because you hear every sentence twice, type it out, and nudge the timing against the picture by hand. The work is not hard, it is just endless, and it starts over with every video. Technical terms, brand names, and proper nouns invite mishears, and every error in the track propagates into every language YouTube auto-translates it into. Underestimate that step, and you spend more time on subtitles than on the actual edit, and still ship a track that gets it wrong exactly where viewers look closest: names, products, numbers.
How does AgenticCutter build subtitles from your video?
It reuses the transcript it already has. AgenticCutter transcribes your video locally with mlx-whisper, and the subtitle track comes straight out of that same text: an SRT file or VTT track whose timestamps match your finished cut exactly, because cut and subtitles are built from one source. You generate subtitles automatically as a byproduct of the edit instead of a separate chore afterwards. Optionally the cutter translates the track into another language while keeping timing and glossary terms intact, and the YouTube assets it produces mean the path from cut to upload is one command, without a detour through a browser upload form.
Which formats and languages do you get?
SRT or VTT, next to your video as a sidecar file, in German or English. If you publish in a second language, the cutter translates the track on request and keeps the timing untouched, so a translated line still sits under the sentence it belongs to. Glossary terms survive translation as well: your product name stays your product name, in every language and every video.
How do brand names and proper nouns stay spelled right?
Through the glossary: you set your recurring terms once, and every transcript and every subtitle track after that spells them the way you wrote them. Product names, guest names, technical vocabulary: maintained in one place instead of corrected in every video. And before anything ships, you see the track and fix whatever is off, because the final say over your subtitles stays with you.
Where does the transcription run, and what leaves your Mac?
On your Mac. mlx-whisper transcribes on-device, and the SRT or VTT file is written straight to your disk, without upload. Your video is never uploaded. If you pick a cloud cut brain, single still frames and transcript text go there. Fully local works with LM Studio.
Can the cutter burn animated karaoke captions into the picture?
Not yet. Word-by-word karaoke captions are planned, not built. Text, overlays, and lower thirds sit one step ahead: they render, but the graphical editor for placing them is still in progress. Today you get a clean sidecar subtitle track, not burned-in caption graphics.
Frequently asked questions about automatic subtitles
Which format do the subtitles come in?
An SRT file or a VTT track, saved as a sidecar next to your video, with timing that matches your finished cut, because subtitles and edit come from the same transcript.
Can the track be translated into another language?
Yes. The cutter translates the track on request into another language, keeps the timing intact, and glossary terms stay exactly as you defined them in every target language.
What if my product name keeps getting misspelled?
That is what the glossary is for: maintain the correct spelling once, and it applies to every transcript and every subtitle track from then on, in every video you cut.
Can I publish straight to YouTube?
Yes. AgenticCutter produces YouTube assets and supports direct upload, so the path from finished cut to published video with subtitles is one command instead of a browser session.
Do I get word-by-word captions like in Shorts?
Not yet. Karaoke captions are planned, text and overlays are in progress. Today you get a clean sidecar track in SRT or VTT, not burned-in caption graphics.
Related features
- Subtitles (SRT/VTT) and translation
- YouTube assets and direct upload
- Agentic via CLI (Claude or Codex)
Related use case: Remove filler words