Dance tutorial subtitles. Keep directions in step.

Start with the lesson you are allowed to share

Polyglot can transcribe the spoken instruction in a recorded dance tutorial locally on an Apple Silicon Mac, help you review or translate it, and export SRT or WebVTT. Use macOS 14.4 or later. The useful result is a checked subtitle track for a specific teaching view: a front-facing demonstration, a rear view or a deliberately mirrored version.

Work from a recording you own or are authorized to process and distribute. Check permission for the instructor, participants, choreography, music and any translated publication. Access to a paid class does not give you permission to download and redistribute it. Resolve excluded people or private studio details in the final video before preparing public text.

Write down which left and right the lesson uses

For each camera change or teaching pass, note the video time, front or rear view, whether the picture is mirrored, the words heard and whose body those words describe. Screen-left is not automatically the learner’s left. Ask the instructor to state the intended convention; a caption editor should not infer it from a flipped picture.

STEEZY’s public site describes front/back views, mirroring, looping and speed control. Those choices illustrate why dance playback needs a view-specific check. They are features of that service, not Polyglot integrations or permission to reuse its lessons. For your own video, keep a small review sheet beside the project.

The fictional example below shows a review decision, not a dance instruction or a tested app result.

Fictional view review
Cut: studio-A-v03
00:24 / front view
Picture: mirrored
Heard: “Start with your
right foot.”

Keep “your right”.
Do not replace it with
“left” just to match
the side of the screen.

Instructor checks the
intended learner action.
Conflict unresolved?
Hold this version.

Keep the teaching count attached to its pass

Review spoken count-ins, pickups, restarts and corrections against the audio. “Five, six, seven, eight” may introduce a demonstration rather than label the next four movements. If the instructor repeats a phrase slowly and then at tempo, retain the count that was actually spoken in each pass. Do not copy one pass’s subtitle timing onto another.

Listen for short words such as “and”, “again” and “other side”. A cleaner-looking sequence can lose a pickup or correction. Music may mask speech or produce incorrect automatic text; remove invented words only after listening and have the instructor resolve uncertainty. Polyglot does not detect beats, recognize steps or evaluate choreography.

If the audio says one direction and the picture demonstrates another, return the conflict to the instructor. Correct the recording or provide an instructor-approved clarification in the lesson itself, then review the changed cut. Captions should not silently replace the spoken instruction.

Review what speech alone leaves out

W3C distinguishes captions for speech and relevant sounds from descriptions of essential visual information. A silent hand placement, weight transfer or change of facing is not recovered by transcribing the soundtrack. Plan the additional description with the instructor; do not turn an uncertain visual interpretation into a subtitle.

Have a person check meaningful music changes and audible cues as part of caption finishing. Translated tracks need a competent language reviewer who understands the instructor’s directional convention. Preserve body-relative wording and the order of spoken counts; put agreed terminology in the lesson notes. A valid timed file does not prove that the movement is understandable or correct.

Accept the picture, language and player together

Export a separate reviewed SRT or WebVTT for the approved final cut and language. For a creator-owned YouTube lesson, its help documentation explains uploading a supported subtitle file with timing. Other players need their own format check. Keep the cut identifier, view or mirror state, subtitle language and unresolved review items together in the handoff.

Watch from the learner’s position with captions on. Check the opening convention, camera switches, repeated demonstrations and the last cue before each restart. At phone size, make sure text does not hide the feet, hands or body position being discussed. Positioning depends on the destination player; a plain subtitle file does not guarantee a safe display area.

If the receiving player offers mirroring, alternate camera views, slower playback or section loops, test each intended mode. Check whether controls affect only the picture or also the caption display, and whether subtitles reappear correctly at loop boundaries. A video file edited to change speed or shot timing needs a new timing review; player speed control is a different case.

Hold a mode with unresolved audio/picture conflicts. Deliver only the approved combinations of video, view and language. Polyglot prepares local subtitle files; it does not mirror the lesson, generate movement descriptions, burn captions into the picture or publish to a dance platform.

Keep exploring.