Muse MuseConcept

Independent concept built on Meta's Model API. Not an official Meta product.

Class Capture Notes

Status: Live.

  1. Your muse

    Rebuild ofTurn a recorded class or meeting into notes that show who said what (opens in a new tab)

    Turn a recorded class or meeting into notes that show who said what. This is the first demo for muse-voice-transcribe-1.0. No video for Muse Voice Transcribe on the first page.

  2. Done

    Checks passed

  3. The plan: Class Capture Notes

    Turn class or meeting recordings into labeled notes

    1. Upload a recorded class or meeting audio file
    2. Let the app convert it in the browser and transcribe it
    3. Label the estimated speaker turns with real names
    4. View, edit and copy the structured notes with summary and actions
    See the full plan

    Who it's for

    Students, teachers and small teams who record classes or meetings

    How it works

    1. Upload a recorded class or meeting audio file
    2. Let the app convert it in the browser and transcribe it
    3. Label the estimated speaker turns with real names
    4. View, edit and copy the structured notes with summary and actions

    Screens

    • Record upload. Upload audio, show conversion and transcription progress.
    • Notes editor. Show transcript with editable speaker turns plus summary and actions.

    What goes in

    • Audio: Recorded MP3, M4A or WAV of a class or meeting up to 10 minutes. App converts to mono 16-bit WAV at 16 kHz in the browser.

    What comes out

    • Text: Plain transcript with editable speaker turns.
    • Text: Structured notes with summary, key points and action items ready to copy.

    Models

    ModelUsed forWhy
    Voice TranscribeTranscribe the recorded class or meeting audioConverts the converted WAV file into a plain transcript with turn timing.
    Muse Spark 1.3Format transcript into readable notesSegments plain transcript into estimated speaker turns and writes summary, key points and action items.

    Limits

    • Rebuild fit: clean. Rebuilt from a text description without watching the original video so layout and script are new, and speaker names are user edited because the current model returns plain text only.
    • Transcripts have no true speaker IDs and no word timestamps. Speaker turns are estimated from text and user labels.
    • Works on recorded files only, not live transcription or live streams.
    • Large files upload to temporary storage first and are deleted after 24 hours. Requests to the app server are capped at about 4.5 MB.
    • Notes are saved only in the browser localStorage.

    From your inspo

    Rebuild of the video titled First muse-voice-transcribe-1.0 demo. It keeps the core demo of turning a recorded class or meeting into notes that show who said what, rebuilt on current models.

    Takes

    • Turning a recorded class or meeting into notes that show who said what

    Leaves out

    • The original video footage and its script
    • The Meta and Llama names and logos
    • Any exact visual design from the developer video

    Cost per run: about $0.02 (est.). 5 minute recording at $0.18 per audio hour is $0.015 plus Spark with about 3000 input tokens at $1.25 per million is $0.00375 and about 1000 output tokens at $4.25 per million is $0.00425 for a total of $0.023.

  4. Recorded demo
  5. Done. Class Capture Notes is built and verified.

    7 m 51 s · Input 4M · Output 42.4k · $2.59 est.