Narrated slide captions. Keep every slide covered.

Caption the recording that belongs to each slide

A recorded lesson or briefing can contain a narration clip on each slide, an embedded demonstration and a longer audio file that continues while the slides change. Those arrangements need different caption checks. A transcript of the whole talk is useful reading material, but it does not automatically attach timed text to the right media objects in the presentation.

Polyglot helps you transcribe authorized audio and video locally on an Apple Silicon Mac running macOS 14.4 or later, review the words and export WebVTT or SRT. Start with the actual recordings, not the PowerPoint file: Polyglot does not open decks, extract slide text or attach captions inside PowerPoint. This guide covers prerecorded narration; live presentation subtitles are a separate PowerPoint feature.

Confirm the PowerPoint version and delivery path

Microsoft documents WebVTT caption support in PowerPoint for Mac from version 16.63, audio caption insertion from 16.71 and UTF-8 SRT insertion from 16.91. Check the version and controls available to the recipient. WebVTT is a practical choice for a compatible installation, but neither a file extension nor a successful local export proves playback in every PowerPoint edition.

Decide whether viewers will receive an editable deck, a presentation link or an exported movie. Record that decision alongside the PowerPoint version, media revisions and caption languages. Microsoft also offers caption generation and translation for embedded videos in PowerPoint for the web. Polyglot provides a local preparation route; it is not required for every PowerPoint caption task and does not integrate with the presentation editor.

Use recordings, slides and included third-party material you own or are authorized to process, translate and share. Check speaker and institutional AI permissions. A private deck can contain attendee names, internal screenshots or licensed teaching material; local transcription does not authorize a wider audience.

Save the actual media and keep its own clock

Use the original recording files where available. Microsoft documents Save Media as for embedded audio and video; on Mac, that feature requires Microsoft 365 and version 16.24 or later. Control-click the media object and look for the command. If it is unavailable, obtain the original file from the author instead of assuming Polyglot can extract it. Keep the saved media unchanged while preparing captions.

Create a small manual map with the deck revision, a stable slide heading, the media filename and each language track. A slide number alone is fragile when someone inserts a new slide. If one slide has two media objects, list them separately. Preserve the original recordings and existing caption files before replacing anything.

Transcribe each clip in a supported recognition language, or import its reviewed matching subtitle file. Review names, numbers, abbreviations and references such as “the lower chart” against the recording and slide. Correct recognition mistakes; ask the author to resolve actual spoken errors. Slide text is a spelling reference, not a substitute for what was said.

A cue ten seconds into a slide’s recording belongs at ten seconds in that media file. Do not add the duration of the preceding slides. Translate after reviewing the source, then have a competent reviewer check meaning and reading time. Speech recognition supports 25 European languages; translation choices do not expand recognition coverage.

Resolve audio that continues across slides

Microsoft explicitly warns that audio captions display only on the slide containing the audio object, even when Play Across Slides lets the audio continue elsewhere. A correct WebVTT attached to the opening slide therefore does not establish caption coverage for the rest of that narration.

For a deck that viewers navigate, ask the presentation author to restructure the narration into slide-specific recordings where appropriate, then prepare and attach captions against each resulting recording. Recheck starts, ends and any speech crossing the split. Polyglot does not split a deck or move its media objects. Simply copying the same full-length track onto later slides will not solve the change of playback clock.

If continuous playback is the intended experience, the author may instead prepare a final movie and distribute it through a player that accepts captions. That changes the deliverable: verify the movie’s audio and visuals first, then prepare its full-length track. The decision belongs to the author and receiving audience, not to an automatic caption repair.

This fictional review note makes the failure visible before release. It is a manually prepared checklist, not an app-generated inspection.

Fictional deck: induction-r4
Slide: Equipment overview
Audio: overview-r2.m4a
Track: overview-r2.en.vtt

00:24: advance to Care.
Audio continues.
Captions disappear.
Author creates separate
Overview and Care recordings.

Check each new recording:
- first word and last word
- speech at the split
- caption selection on replay
Recipient check: pending

Insert and replace tracks without losing other languages

Export WebVTT, or UTF-8 SRT when your receiving version supports it. In the documented Mac workflow, select the media object and use Playback → Insert Captions. Choose the file that belongs to that exact recording, then play it. Naming an SRT file with a .vtt extension does not convert it.

Microsoft’s Mac instructions describe removing inserted captions before editing and reinserting the file. Remove All Captions removes every caption file assigned to that media object, including other languages. Keep all approved originals outside the deck and reinsert the full intended language set after replacement. Do not overwrite an approved translation just because one source cue changed.

In the recipient’s presentation view, select captions and play each media object. Check the beginning, a distinctive middle phrase and the ending, then pause, replay and advance to the next slide. Return directly to a later slide and start its recording. Confirm the correct language and text appear without needing to play the deck from the beginning. Check that captions do not cover essential figures or instructions.

Treat an exported movie as another caption delivery

Microsoft’s Mac export guidance distinguishes recorded Slide Show narration from other embedded media and warns about limitations in the exported movie. Before captioning, watch the actual exported file with sound: confirm every intended recording, slide transition and visual explanation survived. Do not infer export behavior from how the editable deck plays.

Prepare the movie’s captions against its final continuous timeline. Separate slide tracks cannot simply be renamed into one movie track; transitions, pauses and export timings affect where speech occurs. An earlier full-talk transcript may help review wording, but its timestamps still need verification against this exact file.

Upload or attach the timed file using the chosen video player’s supported workflow. Check caption selection, seeking across slide changes, the first and last cue and every language in the audience-facing view. Do not assume captions inserted into the deck are burned into, or automatically carried by, its exported movie. No PowerPoint installation, export or recipient upload was tested for this guide; these are documented boundaries and proposed acceptance checks.

W3C’s caption guidance calls for accurate speech and meaningful non-speech audio. Review those elements manually and provide any needed description of visual-only information separately. Captions alone do not establish the accessibility of the slides or navigation. Keep the approved deck, recordings and tracks together, and repeat the affected checks whenever narration or slide order changes.

Keep exploring.