Generate and import captions

Generate from speech#

1. Open Captions → Generate / import.
2. Choose the language and words-per-caption setting.
3. Start generation and allow the selected transcription model to download on first use.
4. Review the resulting cues in Edit cues, checking names, punctuation, timing, and quiet speech.

Recognition runs locally after the model is available. The initial model download needs an internet connection and disk space. Generation accuracy depends on the recording and model; manually review captions before publishing.

Import existing subtitles#

Import accepts SRT, VTT, and ASS. VTT import retains cue text and time ranges, removes markup, and ignores placement and region metadata as well as invalid cues. Imported subtitle presentation is not a promise of exact visual reproduction of another application's styling.

Make the words readable#

Select the text track in Edit cues. Select a cue and refine its appearance in Inspector, or use Titles → Style selected to apply a consistent preset. Preview captions against the actual footage and keep line lengths comfortable for the intended output.

If generation stalls or fails#

Check connectivity during the first download, available storage, and whether the source has usable speech audio. Try a short section or a smaller supported model if available. If recognition is unsuitable, import a prepared subtitle file and continue editing. See Edit and export caption cues for corrections.