Lab protocol captions. Keep the two clocks clear.

Caption the demonstration, keep the protocol authoritative

A laboratory teaching video often compresses a long wait, repeats a close-up or shows only one step of a larger method. Its captions must follow the edited soundtrack without making the demonstration appear to be a complete real-time procedure. Polyglot can transcribe authorized recordings, translate reviewed text locally on an Apple Silicon Mac and export separate SRT or WebVTT files. The protocol owner still decides whether the recorded instruction is correct.

This workflow is for a research group or educator preparing its own approved demonstration for a compatible video host. It is not scientific validation, an experiment record or a substitute for the complete approved protocol and its safety requirements. Confirm the host accepts timed caption files before starting; permission to attach a movie to a paper does not establish caption playback support.

Identify the step and revision before transcribing

Record the protocol title, exact version or dated identifier, demonstrated step, video filename and intended host in a separate review sheet. Note which steps are omitted and where the edit compresses, repeats or rearranges action. A viewer should be able to find the procedure that actually matches the demonstration, rather than a newer revision with different instructions.

Protocols.io documents videos inside individual steps and a versioning workflow that preserves the earlier published record. Those are useful reasons to check step references when a procedure changes. They do not prove that its embedded player accepts a separate caption track. Keep the protocol reference and the confirmed video-caption destination explicit.

Obtain permission to process the recording and distribute the intended languages. Review contributors, institutional restrictions, unpublished work and visible labels before making the public cut. Keep restricted originals and private review notes separate. Resolve any removal in the audio or picture first; deleting a caption does not remove what the recording reveals.

Use playback time for captions, procedure time for instructions

Generate the source transcript from the finished narrated edit. Subtitle start and end times refer to that media file. A spoken waiting duration belongs in the caption text; it is not a cue duration or evidence that the recording shows the whole wait. Do not stretch a caption across a compressed sequence merely to imitate elapsed experiment time.

In this fictional timing example, a protocol specifies a twenty-minute wait and the edited video shows six seconds of that interval. The reviewer checks the narration against the approved step and ensures the video explains the compression. The numbers illustrate an editorial check, not instructions for an experiment or an app-test result.

Fictional review sheet
Protocol: Demo-A, revision 3
Step: 7
Video: demo-a-r3-cut02.mp4

Video 00:42–00:45
Heard: “Wait twenty minutes.”
Caption follows that speech.

Video 00:45–00:51
Compressed waiting sequence.
Six seconds of screen time
is NOT the procedure duration.

Owner check:
Does the video explain the cut?
Does step 7 match revision 3?

Decide which kind of discrepancy you found

If recognition turns an audible “twenty” into “two”, correct the subtitle to match the speech. If the narrator says “two” while the approved protocol says “twenty”, hold the video for the owner: changing only the caption would conceal a conflict. The owner can revise the narration or provide an appropriate explicit correction, then review the new final edit and its captions together.

Check units, decimal marks, abbreviations, sample labels and step numbers against both the audio and the approved reference. Do not infer an inaudible quantity from a nearby label. Repeated close-ups need another listening pass because the narration may explain a different stage even when the picture looks familiar.

W3C caption guidance includes meaningful non-speech audio as well as speech. Review relevant sounds with suitable caption tools and check any essential silent visual action separately. Speech recognition cannot describe an unspoken technique, identify an instrument signal or establish that the demonstrated procedure is safe.

Translate after the source and procedure owner agree

Have a reviewer competent in the target language and the subject check duration words, units, negation, conditional steps and abbreviations. Keep approved identifiers stable where the owner requires them. A fluent translation that changes “until” to “for”, or drops a condition, can describe a different instruction.

Polyglot preserves source timing during translation and validates cue structure; these checks do not validate scientific meaning. Review source corrections in every affected translation. If a new protocol revision changes the demonstration, preserve the older approved files and create a clearly identified replacement package rather than silently relabeling the old captions.

Accept the video, caption track and protocol reference together

For an institution-approved YouTube teaching upload, YouTube documents attaching a subtitle file with timing and selecting its language. Use the accepted file format and check the actual hosted video. Other repositories, learning systems and journal portals need their own verified delivery method; a downloadable SRT beside a movie does not guarantee that the player will display it.

Play each language from the start, seek to the demonstrated step, cross every compressed interval and inspect the final cue. Check that captions clear when the narration stops and do not cover a scale, control or action the learner must see. Open the protocol reference and confirm its revision and step still match. Keep a readable transcript if required, but label it as text accompanying the demonstration, not the authoritative method.

After replacing either the video or the protocol reference, repeat those checks and remove stale public caption attachments as appropriate. Release only when the owner accepts the recorded explanation, the language reviewer accepts its wording, and the receiving player displays the correct track. Polyglot creates the local subtitle files; uploading, version mapping and this acceptance review remain your work.

Keep exploring.