Made for your workflow
Video figure captions. Keep the review copy anonymous.
A research video and its subtitles travel together. Check the anonymous review package, then prepare a separate caption track for the credited publication version.
Apple Silicon · macOS 14.4 or later
01
Approve the review cut
Resolve identifying speech and pictures in the video before creating its captions.
02
Check every attachment
Review timed text, filenames and the recipient’s permitted delivery format.
03
Separate the public release
Pair later credits and narration changes with a newly checked caption version.

Caption the video figure that reviewers will actually receive
A video figure is a supplementary research recording that demonstrates part of a paper: an interaction, a prototype or a result that is easier to show in motion. When the submission uses anonymous peer review, the subtitle attachment needs the same identity check as the video. Polyglot can create, review and translate timed subtitles locally on an Apple Silicon Mac, then export a separate SRT or WebVTT file.
SIGCHI’s video guide describes video figures separately from previews and recorded talks, and calls for timed closed captions. Venue instructions take precedence: for example, the published CHI 2026 video guide specifies SRT or SBV, while its papers instructions require anonymized supplementary material during review. These are dated examples, not confirmation of a current submission window or another venue’s rules. Check your actual track, year and submission portal before exporting.
Use media your team is authorized to process and submit. Confirm participant consent covers the intended review or public release, and clear any third-party footage, music and interface content. Anonymizing authors does not by itself protect research participants. Keep confidential material within the access arrangements approved for the study.
Resolve identity in the recording before editing the words
Work from a dedicated review copy with a neutral filename. Have the responsible author inspect spoken affiliations, title frames, institutional marks, author faces, desktop paths and visible account details. SIGCHI’s guidance explicitly addresses author and affiliation information when anonymization is required. Video editing, face treatment and media metadata inspection happen in suitable external tools; Polyglot does not anonymize recordings or certify that an identity cannot be inferred.
Create captions from the approved review cut, or import a matching existing SRT or WebVTT. Listen to every line. If the audio still identifies the lab, deleting its name from the subtitle leaves an identifying recording and an inaccurate caption. Return that issue to the author and video editor. Once the audio has been appropriately revised under the venue’s rules, review captions against that revised sound.
Keep a private issue sheet outside the submission package. Distinguish an identification spoken in the recording from a name accidentally introduced by recognition, an obsolete caption from an earlier edit, and an identifier present only in a filename. Each has a different fix. Do not remove scientific attribution or required citations merely because they contain a person’s name; ask the responsible author to apply the venue’s policy.
Review the caption file as a separate disclosure surface
This fictional example shows why the final media, subtitle and attachment names need separate decisions. It is a manual review note, not a Polyglot detection result. The neutral filename is illustrative; use any naming convention required by your submission portal.
After resolving issues, inspect the exported text as well as playback. Check cue wording, speaker labels, project references, URLs and any notes added by a previous editor. Search for known identifiers as an extra pass, then read the whole file: a search list cannot establish anonymity. Review filenames and any surrounding archive or README separately, without including the private identity-check sheet.
Keep captions faithful to what can be heard. Add meaningful sound information and necessary speaker context after human review; use neutral labels when the venue permits them. Polyglot does not automatically identify speakers or detect disclosure risks. Captions also need enough time to read and must not cover the interface detail or physical action the figure demonstrates.
Fictional private note
Review cut: figure-r03.mp4
00:08: lab name still spoken
Fix: author revises video
00:32: caption has old URL
Fix: check final audio
File: named-lab-demo.srt
Fix: permitted neutral nameKeep the accepted publication version separate
If the paper is accepted and the venue requests a credited version, preserve the reviewed anonymous deliverables in their existing restricted location. Create a clearly separate publication package. A new title card, acknowledgement, voiceover or end credit can change both the wording and the time at which later cues appear.
Do not attach the anonymous subtitle file to a credited cut merely because most of the demonstration is unchanged. Compare the opening, each changed passage and the ending, then watch the full new version. A constant timing shift only fits a genuinely uniform offset; inserted or replaced sections require their own review. Retain approved wording where appropriate without silently replacing completed translations or losing manual edits.
Prepare only the languages the recipient accepts. When translating, ask a language reviewer to check that affiliation terms, project names and speaker labels remain appropriate for that particular version. A fluent translation can reintroduce an old identifying phrase if it was made from an earlier source. Keep source version and review status visible in your private handoff notes.
Check the submitted pair, not just the local draft
Export the timed format explicitly accepted by the destination. Polyglot exports SRT and WebVTT; it does not export SBV. Where the current instructions accept SRT, that is a supported option. A readable text transcript is not a substitute for a required timed file, and captions burned into the picture are a different deliverable. Upload and submission happen outside Polyglot.
Open the exported track with the exact final video in a compatible player. Check the first spoken phrase, a distinctive middle line and the last cue, then review the entire recording. Verify cue order, text, accented characters and language selection. At small player sizes, confirm that captions do not hide the interaction being demonstrated. Review sound information and essential visual explanation with the appropriate accessibility reviewer; a successful export is not an accessibility certificate.
Where the portal allows it, inspect or download the actual submitted attachments and repeat the pairing check. Confirm the intended review or public version, permitted filenames, duration, language and caption format. Keep a private receipt of the submitted pair and replace both files together whenever a changed video requires changed captions.
Polyglot supports Apple Silicon Macs running macOS 14.4 or later. Media and subtitle processing stay local; requested downloads, user-controlled updates, default-on content-free analytics with an opt-out and Pro licensing have separate network boundaries. New AI subtitle jobs use a seven-day trial and require Pro afterward. Editing and exports remain free.
Sources & further reading
SIGCHI: video categories, caption files and anonymous author information
CHI 2026: video submission and SRT or SBV captions
CHI 2026: anonymous supplementary material during paper review
W3C: captions and human review
Polyglot: privacy and network boundaries
Published by Polyglot using an AI-assisted editorial workflow. How we prepare and update our guides.
Keep exploring.
Try it on your Mac.
Find the right tools for your words and video, with processing on your Mac.
Download nowApple Silicon · macOS 14.4 or later