Automatic captions.
Editable on your Mac.

1. Choose the speech to caption.

This workflow is for Vibecut 0.4.3 developer preview on an Apple silicon Mac with macOS 12 or later. Account access, an internet connection, and available Creative Credits are required for the cloud caption task. Downloading the app does not add credits. Follow the getting-started guide to install and import your recording.

Listen to a short selection and make sure it contains the speech you want to caption. Check for music masking the voice, overlapping speakers, or sections where no speech is audible. Keep your original source files available.

Finish the main pacing decisions first. If you plan to remove long silent pauses, do that pass before caption review. Later changes to the edit can require another timing check.

2. Run Smart Captions.

Choose the audio selection and use Smart Captions in the right-side agent area. Check the target material and the Creative Credit quote before confirming. Caption generation is its own cloud task; the per-minute prices for Analyze are not a caption price.

The result is timed text clips on your timeline. Treat the generated wording, sentence breaks, and timestamps as a draft to inspect. Recognition can be wrong even when the audio sounds clear to you.

A goal for your review pass:

“Keep each caption to a readable phrase. Check names and technical terms, and match the text to the spoken sentence.”

Check pricing and how credits work for account options. Use the quote for the selected task. Undoing text in a project does not refund completed cloud generation.

3. Correct one sentence at a time.

Play the speech while reading the caption, then inspect its text clip on the timeline. The text is editable: you can correct wording, split or adjust sentences, and refine timing.

1. Listen for the actual words. Check names, numbers, technical terms, and words spoken quietly. Preserve what the speaker meant when adding punctuation.
2. Check the sentence break. If a caption ends halfway through a thought or stays on screen as a long block, split or adjust it into readable phrases.
3. Check the start. Move through the relevant timeline frames and compare the text’s appearance with the beginning of the spoken phrase.
4. Check the end. Make sure the caption has enough time to be read and does not linger into a different sentence or scene.
5. Play through the boundary. Read the previous and next captions together. Look for missing words, repeated wording, and unexpected gaps or overlaps.

Illustrative correction: a speaker says “We kept the pause. It makes the answer feel honest.” If both sentences appear as one unwieldy caption, split the text at the sentence break and adjust each text clip to the speech.

Use Undo or edit history if a correction makes the sequence worse. After changing video cuts, replay the affected captions against the new timeline.

4. Make the text readable in the final frame.

Choose a clear font and a size that remains readable in the intended viewing format. Adjust caption position so the text does not cover a face or important on-screen information. Inspect the actual composition with its titles, stickers, and effects.

A text outline in a contrasting color can help captions stay readable as the background changes. Review both bright and dark moments; a style that looks good on one frame may disappear on another. Keep enough space around the text to avoid clipping at the edge of the picture.

Export a short test from the Mac app and open the actual output in a separate player. Watch it with sound to check timing, then without sound to check whether the captions are understandable. This guide covers captions displayed in the video; it does not promise a separate subtitle-file export.

When captions need another pass.

The text is wrong.

Check the selected source and listen to the original speech. Correct isolated errors on the timeline. Music, accents, specialist vocabulary, and overlapping speakers can affect recognition. A short selection with clear speech is easier to inspect than regenerating an entire episode repeatedly.

The wording is right, but the timing is wrong.

Compare the affected text clips with the spoken phrases and adjust their timing. Check whether the recording has been trimmed, moved, or otherwise edited since generation. Inspect neighboring clips so correcting one caption does not create a new overlap.

The task reports unusable captions or timestamps.

Preserve the exact error and request ID. Confirm that the selection contains audible speech, check your connection and account access, and try a short, clear sample. If it fails again, contact support with the app version, selected source, and approximate timestamp. A repeated failure is worth investigating before spending time on more attempts.

Email davidbinghe@gmail.com if needed. Send only media you have permission to share; passwords and complete payment details are never needed.

Common questions.

Do I need to analyze the entire video first?

Smart Captions is a separate task with its own quote. A full visual analysis is not a prerequisite for basic property edits. Use Analyze when an editing request needs information about what happens in the source; see the podcast guide for choosing Audio only or Video + audio.

Can I use languages other than English?

Editing instructions do not have to be written in English. Speech recognition quality still depends on the language, accent, and audio. Try a short representative sample and review it before committing to a longer recording; this guide does not guarantee support or equal accuracy for every language.

Will captions identify every speaker?

This workflow does not promise reliable speaker identification or automatic camera switching. Review dialogue with overlapping speakers carefully, and check that any wording or labels you add represent the recording accurately.