Word-highlight captions from a transcript made on your PC

Captions help viewers follow a short even with the sound off, and a caption that highlights each word as it is spoken keeps their eyes on the screen. The hard part is the timing: re-timing captions word by word by hand is one of the slowest jobs in short-form editing.

CoClip builds those captions from a word-level transcript, and by default that transcript is made right on your computer.

Two ways to transcribe

When you choose Generate transcript, you pick the engine:

• This PC (Whisper). Speech recognition runs on your computer, with no account and no key. It works offline and understands Vietnamese and English. On your first transcription CoClip asks before downloading a speech model once: Best quality (574 MB) or Fast (264 MB).
• ElevenLabs Scribe. If you prefer a cloud engine, add your own ElevenLabs API key in Settings → Connectors. The audio goes straight from your computer to ElevenLabs, and ElevenLabs bills your account under its own pricing. CoClip asks before a long file uses your credits.

There is no CoClip cloud in between either way. CoClip doesn't resell transcription, your key is stored encrypted on your computer, and your projects stay as plain folders on your machine.

From audio to captions in three steps

1. Pick the audio

Select the voice track, or any clip, and choose Generate transcript.

2. Choose the engine and language

Transcribe on this PC or with ElevenLabs Scribe, in Vietnamese, English or with the language detected for you. The word-level transcript lands on your timeline.

3. Edit the transcript

Fix a name, correct a misheard word or tidy the punctuation in the transcript. The captions update with it, so you never have to re-time a caption by hand after a correction.

Prefer to transcribe elsewhere? You can also import a transcript and generate captions from that.

Captions that look like yours

Word-highlight captions are only the starting point. You can style them the way your channel looks:

• rounded text boxes,
• stroke and shadow,
• a highlight colour or box for the active word, and emphasis for key words,
• reusable styles, so the next video starts with the same look.

Captions come in short pages of a few words that also break on natural pauses, and you can adjust the text or timing of any single word.

Text and captions are edited right on the preview: click to select, drag with smart guides, and double-click any text to edit it in place.

Vietnamese captions, rendered correctly

If you publish in Vietnamese, you know how often captions break on diacritics. In CoClip, text and captions render Vietnamese diacritics correctly, including stacked marks like Ở and Ữ.

The on-device engine handles Vietnamese too, so a Vietnamese voice-over can go from audio to styled captions without leaving your computer.

What you need

• CoClip on Windows 11 or macOS. On-device transcription is included in the free version.
• A one-time download of the speech model, on your first transcription. After that, transcription works offline.
• Optional: an ElevenLabs API key, if you want Scribe. The same key also turns on AI voice-overs and music.

Working with your AI agent

Captions also fit into agent editing. If you connect Claude, Codex or Gemini through CoClip's built-in MCP server, you can ask your agent to transcribe the voice track, add captions or restyle them, and the change appears live on the timeline, labelled in the AI log and one click from undo.

Get started

Download CoClip, choose Generate transcript on your next voice-over, and let it run on your PC. The free version includes captions and transcription; the pricing page shows what Pro adds.