Caption Generator
Transcribe your timeline on your own machine, then drop styled, word-by-word animated captions — no API key, no per-minute fees.
What it does
Caption Generator transcribes your timeline with AI speech recognition that runs entirely on your machine by default (OpenAI's Whisper models) — your audio stays local, there's no API key to paste, and no per-minute fees. Pick a model from fastest to most accurate, enable GPU acceleration if you have one, and transcribe the whole timeline or just your In/Out range.
Prefer the cloud? If you'd rather not download local models — or you want a particular engine's accuracy — bring your own key and pick Deepgram Nova-3, ElevenLabs Scribe, or Speechmatics from the same model menu. Cloud engines send your audio to the provider you choose and bill on your own account (BYOK), while everything after transcription — the review editor, styles, and native MOGRTs — works exactly the same.
The transcript lands in a full review editor — search & replace, speaker labels, line-length and case reformatting, even a word censor — before a single caption is placed. Then choose a style from the built-in library (including word-by-word karaoke highlight looks) or load your own custom caption MOGRT, and PrimCut places every caption as a native, editable Motion Graphics Template on your timeline. Prefer files? Import and export SRT/VTT.
Key features
- Local by default — built-in Whisper models transcribe on your machine; no key, no account, no per-minute cost.
- Optional cloud engines — bring your own key for Deepgram Nova-3, ElevenLabs Scribe, or Speechmatics when you want their accuracy or want to skip local model downloads.
- Word-by-word karaoke styles — the active word highlights as it's spoken, like the biggest creators use.
- Your own caption MOGRTs — keep your brand's look; PrimCut fills your template with synced text.
- Full review editor — search & replace, speaker labels, reformat line width and case, close gaps, censor words.
- SRT / VTT import & export — bring captions in from anywhere, or hand them off to any platform.
- Model choice + GPU acceleration — from a fast first pass to maximum accuracy, with translation available when you need it.
How to use it
Transcribe the timeline
Pick your audio tracks, language and model — local Whisper or a cloud engine — or just hit go. In/Out-only and speaker labels are right there in the same dialog.
Review the transcript
Fix names with search & replace, tidy line lengths in the reformat tool, and follow along with the playhead.
Pick a style
Choose a built-in look — karaoke highlight included — or load your own custom caption MOGRT.
Generate
Every caption lands on your sequence as a native MOGRT, synced and styled — still fully editable in Premiere afterward.
Watch the walkthrough
Frequently asked questions
Is my audio uploaded anywhere?
Not with the default local engine — Whisper transcription runs on your machine with locally installed AI models, and works with the internet off. If you choose one of the optional cloud engines (Deepgram, ElevenLabs, or Speechmatics), your audio is sent to that provider to transcribe — that's the only time it leaves your machine. (The optional translation feature also uses an online service.)
Do I pay per minute of transcription?
Not with local Whisper — because the AI runs on your hardware, an hour-long podcast costs the same as a 30-second short: nothing beyond your PrimCut plan. The optional cloud engines are billed by the provider on your own key (BYOK), so those follow the provider's rate.
Do I need a cloud engine, or is local enough?
Local Whisper is the default and needs no key — it's plenty for most work. The cloud engines (Deepgram Nova-3, ElevenLabs Scribe, Speechmatics) are optional, for when you want a specific engine's accuracy or would rather not download local models.
Can I use my own caption design?
Yes — load any custom caption MOGRT you've made (or bought) and PrimCut fills it with your synced transcript, so your brand look stays intact.
Are the captions editable after generating?
Fully. Each caption is a native Motion Graphics Template on your timeline — retime it, restyle it, or fix a word in Essential Graphics like any other MOGRT.
Related tools
Get Caption Generator in your panel.
Eleven tools, one Premiere Pro panel. Free tier, no credit card.