Skip to content
Vynote — Vynote home

Features

What the current development build of Vynote does — and what it does not do yet. No public build has been released.

What the labels mean

Implemented
Built and verified in the development build.
Experimental
Working in testing, but not yet validated in real-world use.
Planned
Intended, not built.
Not available
Not part of Vynote today.

Live meetings

Experimental

Live meeting transcription

Follow a call while it happens. Vynote listens to what your Mac plays and what you say, and turns both into timestamped lines — on your Mac.

Experimental. Tested with synthetic speech on one Mac, including a 14-minute session. Not yet tested in real Google Meet, Microsoft Teams or Zoom calls.

Vynote Live Meeting: a LIVE session with system audio and microphone capturing, timestamped finalized lines and one line still in progress.
Vynote Live Meeting: a LIVE session with system audio and microphone capturing, timestamped finalized lines and one line still in progress.

Live Meeting in the Vynote Mac app, with demonstration content.

  1. 01

    Live transcription

    Experimental

    Transcribe a meeting on your Mac as it happens.

    • Starts only when you press Start; nothing is captured on launch.
    • Text appears progressively in short rolling windows.
    • Pause finishes speech already captured, then stops capturing; Resume continues on the same timeline.
    • Stop finishes outstanding audio, saves the transcript and opens it for review.
    • Base keeps up with two sources in testing; Small may fall behind.
  2. 02

    Microphone and system audio

    Experimental

    Capture both sides of a call at the same time.

    • System audio: what other participants say, as played by your Mac (Meet, Teams, Zoom…).
    • Microphone: what you say into the default input device.
    • Both are on by default; you can start with just one.
    • If one source fails, Vynote pauses and lets you continue with the other, retry or stop.
    • With speakers instead of headphones, remote speech can be transcribed twice. Use headphones.
  3. 03

    Session review

    Experimental

    Go back to what was said in a live session.

    • Finished and interrupted live sessions appear under Recent live sessions, also after restarting.
    • Review is read-only and text-only: live audio is never saved, so there is nothing to play back.

Recordings

Implemented

Turn recordings into editable transcripts

Bring an audio or video file that is already on your Mac. Get a transcript you can search, correct and navigate by time.

Local file transcription

Implemented

Transcribe recordings already on your Mac.

  • MP4, MOV, MP3, M4A and WAV, up to one hour, when macOS can read the audio.
  • Drag and drop, or Choose a file (⌘O). Files stay where they are and are never modified.
  • The whole recording is processed; you can cancel at any time.
  • Silent recordings complete with “No speech detected”.

Transcript editing and search

Implemented

Fix what was misheard and find what was said.

  • Edit any segment in place; changes are saved locally and survive restarts.
  • Search across the transcript (⌘F).
  • Transcribing again keeps one previous result that you can restore.
Vynote workspace with media player, transcript segments and timestamps.
Vynote workspace with media player, transcript segments and timestamps.

Screenshot of the Vynote interface with demonstration content.

Everyday details

Supporting capabilities

Timestamps you can navigate, models that run locally, your languages and your privacy.

Implemented

Timestamp navigation

Every segment is tied to the moment it was spoken.

  • Click a timestamp to move the player to that point.
  • Timestamps follow the original recording, including leading silence and gaps.
  • Recognition and timing are approximate, not word-perfect.
Implemented

Offline models

Three multilingual whisper.cpp models, downloaded only when you ask.

  • Tiny 77.7 MB · Base 148.0 MB · Small 487.6 MB.
  • Downloaded from a fixed, revision-pinned source and checked against a SHA-256 hash before use.
  • No model is included in the app or downloaded automatically.
  • Removing a model keeps your transcripts.
Implemented

Language support

Built for English and Bahasa Indonesia.

  • Interface in English and Bahasa Indonesia, in Light or Dark.
  • Spoken language: Auto-detect, English or Indonesian.
  • Accuracy for real human speakers and accents has not been measured yet.
Implemented

Privacy

Designed to work without the internet.

  • Transcription runs on your Mac; audio is never uploaded.
  • No account, analytics or telemetry.
  • Transcripts are stored in your user Library with private file permissions.

Not available yet

These are not part of the current build. Some are planned; none has a date.

  • Windows versionPlanned
  • Exporting transcriptsNot available
  • Speaker names or identificationNot available
  • Echo cancellation for speakersNot available
  • Choosing a specific microphoneNot available
  • Deleting live sessions from the appNot available
  • Summaries or action itemsNot available
  • Importing from video linksNot available