Music lesson subtitles. Keep speech and playing distinct.

Subtitle the instruction around the performance

Polyglot can transcribe the spoken parts of an authorized instrument lesson locally on Mac and help you review and translate timed subtitles. It does not transcribe a performance into notation, identify pitches or judge whether a passage was played correctly. Keep the teacher’s explanation, the musical demonstration and any score reference distinct.

Begin with the finished lesson and the exact score edition or original exercise used by the instructor. Obtain permission for recordings, performances, lyrics and displayed notation as needed. Permission to teach a piece does not by itself establish permission to redistribute a particular published score or someone else’s recording.

Create a speech-and-playing map

For each teaching passage, note when the instructor speaks, when an example is performed and where it points in the lesson material. Record a bar or measure reference only after a person has checked the score. MusicXML’s documentation shows that measures are part of musical structure; subtitle cue numbers and elapsed video seconds are different references.

This fictional example distinguishes a spoken count-in from a performed passage. It contains no copied music and is not an assessment of a real performance.

Illustrative lesson map
Video:
original-study-v02.mp4
Exercise: teacher score B

00:18: “Start at bar five.”
00:21: spoken count-in
00:23–00:31: played example
00:33: teacher explanation

Check bar 5 manually.
Do not turn played notes
into invented spoken lyrics.

Review terminology without inventing musical evidence

Check note names, finger numbers, instrument parts and the difference between a counted beat and a numbered bar. Verify unusual terms with the instructor. When speech overlaps playing, listen carefully rather than choosing the most plausible musical phrase. Leave unresolved wording for a person who can hear and understand it.

DCMP provides guidance on describing meaningful music and sound in captions. Use a suitable human-authored description when the sound carries information; do not claim that a generic label communicates every musical detail being taught. Lyrics, when relevant and authorized, need their own careful review. Polyglot does not recognize instruments or generate verified music descriptions.

Translations need an instructor or musically competent language reviewer. Keep score references aligned and avoid silently converting note-naming conventions in a way that conflicts with the visible lesson material. Put any explanatory mapping in the companion lesson notes. A structurally valid subtitle file does not prove musical correctness.

Check the performance and reading space together

Load the exported SRT or WebVTT in the actual teaching player. Watch the transition from explanation to count-in to playing. Make sure a stale spoken caption does not remain across a performance where it suggests a new instruction. Keep the player’s caption area clear of the hands, fingering diagram or score passage under discussion.

Essential silent fingering or posture information may need a separate description. Do not expect an audio transcript to supply it. Test the receiving player at the size students will use, including whether captions can be selected and whether a passage can be replayed without losing the reference.

Hand over the video revision, approved score or exercise reference, reviewed language tracks and lesson map. If the instructor replaces the played example, inspect the surrounding cue timing and score references again. This workflow supports the spoken explanation; it is not a substitute for notation, instruction or human evaluation of the performance.

Keep exploring.