Meeting Notes Studio

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. Show thinking

3. Checks passed

4. 

   The plan: Meeting Notes Studio

   Turn class and meeting recordings into clear speaker notes

   1. Upload a recorded class or meeting audio file
   2. Add speaker names for turn labels
   3. Run transcription and generate labeled notes
   4. Edit speaker labels and summary then save in the browser

   See the full plan

   Who it's for

   Students and small teams who record classes or meetings

   How it works

   1. Upload a recorded class or meeting audio file
   2. Add speaker names for turn labels
   3. Run transcription and generate labeled notes
   4. Edit speaker labels and summary then save in the browser

   Screens

   • Upload. Upload recording, convert to WAV, and enter speaker names.
   • Notes. Review transcript turns, assign speakers, and read summary and actions.
   • Saved. View and reopen notes saved in the browser.

   What goes in

   • Audio: Recorded class or meeting, short clip under 4.5 MB request size. Browser converts it to mono 16-bit WAV at 16 kHz before upload.
   • Text (optional): Optional speaker names such as Speaker 1 is Maya and Speaker 2 is Lee, plus class or meeting title.

   What comes out

   • Text: Readable notes with labeled speaker turns, summary, and action items.
   • JSON: Structured turns with speaker, text, and timing when available for editing and export.

   Models

   Model	Used for	Why
   Voice Transcribe	Transcribe the recorded class or meeting audio to text	It is the speech to text model for recorded files and returns a plain transcript.
   Muse Spark 1.3	Turn the plain transcript into readable notes with speaker turns	It groups text into turns, applies user supplied speaker names, and writes summaries and action items.

   Limits

   • Rebuild fit: clean. Built from the written description only since the video could not be watched, so layout and script are new but the demo job is the same.
   • Transcription returns one plain transcript with no speaker labels and no word timestamps. The user labels each turn with speaker names.
   • Only recorded files are supported. No live transcription.
   • Audio must be converted in the browser to mono 16-bit WAV at 16 or 24 kHz before sending.
   • Saved notes stay in the browser local storage. There are no accounts and no email or sharing service.

   From your inspo

   Rebuild of the video titled First muse-voice-transcribe-1.0 demo. It keeps the same demo job of turning a recorded class or meeting into notes that show who said what.

   Takes

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

   Leaves out

   • The video footage
   • Its script
   • The Meta and Llama names and logos

   Cost per run: about $0.03 (est.). 10 minute meeting uses about $0.03 of transcription at $0.18 per audio hour plus about $0.002 of Spark reasoning. Total about $0.032 per run.

6. Done. Built Meeting Notes Studio and npm run verify exits 0 (typecheck, build, console, 5 smoke/action/error tests, and manifest all passed).

   It's live.

   musemuse-meeting-notes-studio.vercel.app (opens in a new tab)

   Product pageNew muse

   9 m 52 s · Input 4.1M · Output 44.2k · $2.75 est.