Student feedback subtitles. Keep each comment in context.

Work from the feedback recording you will deliver

Polyglot can turn an authorized recording of a teacher’s spoken feedback into timed subtitles on an Apple Silicon Mac. Review the text with playback, optionally translate the reviewed original, and export SRT, WebVTT or a separate text transcript. This workflow starts with an existing recording; Polyglot does not record your screen, grade work or generate assessment feedback.

Keep three things together before transcription: the final feedback video, the exact submission being discussed and the rubric revision used by the teacher. A comment about page three of a first draft may be misleading beside a revised submission, even when every subtitle is spelled correctly. Give the recording a version identifier and use an institution-approved assignment identifier in filenames rather than unnecessary personal details.

Process student work only when you are authorized to do so under your institution’s rules. Confirm that local AI processing and any translation are permitted for this material. Review visible names, marks, peer submissions, notifications and private browser tabs before creating the deliverable. Local media processing does not determine who can access a file after it is uploaded.

Keep comments attached to the right criterion

Make a short review sheet outside the app with the video time, submission location, rubric criterion and any unresolved wording. Check criterion names against the rubric, then listen for small words that change the judgment: “not yet,” “partly,” “could” and “must.” A suggestion to try another example should not become a new assessment requirement in the subtitles.

Separate transcription errors from a mismatch in the recording itself. If the audio says one criterion name while the highlighted rubric row shows another, ask the teacher which was intended. Correct the recording or provide an explicitly approved clarification through the institution’s feedback process. Do not silently rewrite the captions to conceal the disagreement.

The fictional example below is a review note, not generated feedback or a real student record. It catches both a lost qualification and an ambiguous visual reference.

Fictional feedback review
Recording: feedback-v02
Submission: draft-v01
Rubric: essay-v03

01:16 / page 3, paragraph 2
Criterion: use of evidence
Heard: “Not yet supported.”
Draft: “Supported.”
Action: restore “not yet”.

01:28 / “Move this earlier.”
Pointer is between paragraphs.
Action: ask teacher to clarify;
hold delivery until resolved.

Resolve references that depend on the picture

Play through comments such as “this sentence,” “the second bar” or “the value here” while watching the cursor, highlighted passage or demonstration. Confirm that the intended target is visible when the cue appears. If the pointer moves before the spoken comment, check the surrounding sequence rather than guessing from a single frame.

Captions represent the audible feedback. They do not automatically explain an unlabeled diagram or turn a vague spoken pointer into a precise reference. Where the recording is unclear, ask the teacher to rerecord the passage with a useful reference or approve a separate clarification. If essential visual information needs a description or an accessible companion document, arrange that separately; speech transcription alone does not supply it.

After any cut, replacement passage or change to the displayed submission, review the affected cues and the surrounding transitions again. A subtitle file for the old video is not ready simply because the running time looks similar. Keep the superseded version distinguishable from the file being delivered.

Review each language without changing the assessment

Check that the recorded language is supported for transcription; translation targets are a separate capability. Review the original before making another language track. Keep assignment titles, named criteria, numerical marks and references consistent with the approved source documents, while preserving what the teacher actually said.

Have a competent reviewer compare a translation with the original feedback, especially qualifications, negation and the difference between an optional improvement and a required correction. Keep the authoritative feedback record identified according to institutional policy. A translated subtitle track is a reading aid, not a new grading decision.

Polyglot’s structural checks do not establish that a translation preserves an assessment’s meaning. Hold a track when an essential term or judgment is unresolved. The app does not validate rubric compliance, recognize the work shown on screen or decide which students should receive a recording.

Check captions and access in the student’s player

Export the reviewed file in the format supported by the institution’s receiving player. Upload it through the approved platform’s own controls and pair it with the exact feedback video and language. Polyglot produces separate files; it does not upload to a learning platform, set access permissions or synchronize grades. Panopto’s student-assignment documentation describes a broader platform workflow, including time-stamped comments; those are platform features, not Polyglot capabilities.

Use the platform’s permitted student-view or test-account procedure to check the handoff. Confirm that the intended recipient can open the recording and select the correct captions, and inspect the platform’s access list before delivery. Do not widen access to the whole class just to solve a playback problem. Treat downloadable subtitle and transcript files as part of the same private feedback package.

Watch the opening, a rubric reference, a screen-pointing comment and the final next step. Seek into the middle, check caption timing and clearing, and test at mobile size so captions do not cover the passage being discussed. W3C’s player guidance includes keyboard access and visible focus; check that the destination’s caption controls can be operated without a mouse.

W3C’s caption guidance also includes relevant sounds and speaker identification. Review those needs manually, particularly if the recording includes student speech. Finish missing caption events in a suitable editor; Polyglot does not automatically identify speakers or guarantee complete accessibility captions. Deliver only after the teacher’s meaning, the displayed work and the recipient’s playback agree.

Keep exploring.