Documentation
How to install Vynote, transcribe recordings and live meetings, and keep your data where you expect it.
Vynote has not been publicly released yet. These guides describe the current development build.
Getting started
Vynote needs a Mac with Apple silicon running macOS 14 Sonoma or later. You do not need an account or an internet connection to transcribe — only to download a model the first time.
- Install Vynote and open it from Applications.
- Open Settings (⌘,) and download a model. Base is a good first choice.
- Go to Home (⌘1) and drop a recording, or use Choose a file (⌘O).
- Choose Generate transcript. When it finishes, search, edit and click timestamps to jump around.
- For a call, open Live Meeting (⌘3) and press Start transcription.
Your previous imports and sessions are in Library (⌘2). Built-in samples are silent demonstrations you can explore and edit freely.
Installing Vynote
When a release is available, it is offered as a .dmg disk image on the download page, together with its size and SHA-256 checksum.
- Download the
.dmgand, if you like, verify its checksum withshasum -a 256. - Open the disk image and drag Vynote into Applications.
- Eject the disk image and open Vynote from Applications or Spotlight.
Vynote does not install command-line tools or background services. To uninstall, quit Vynote and move it to the Trash. Your transcripts and models stay in your user Library until you remove them (see Privacy and local storage).
Importing recordings
Vynote accepts MP4, MOV, MP3, M4A and WAV files up to one hour long. A file is accepted only if macOS can actually read its audio; some codecs inside a supported container may still fail.
- Drag files onto the Vynote window, or use Choose a file (⌘O).
- Vynote keeps a reference to the original file. It never copies, moves, changes or uploads it.
- If you move or rename the file later, Vynote shows that it can no longer find it.
- Video files are transcribed from their audio. Multiple audio tracks are mixed together.
To transcribe, open the recording and choose Generate transcript. Pick the spoken language (Auto-detect, English or Indonesian) and the model. Only one transcription runs at a time, and you can cancel it.
Downloading transcription models
Vynote does not include a speech model. Download one in Settings before your first transcription. Downloads start only when you choose them.
| Model | Download | Best for |
|---|---|---|
| Tiny | 77.7 MB | Quick drafts; the fastest, with the most recognition errors |
| Base | 148.0 MB | Everyday use and live meetings (recommended) |
| Small | 487.6 MB | Better recognition for files; may fall behind in live meetings |
- Models come from a fixed, revision-pinned address on Hugging Face and are checked against a known SHA-256 hash. A file that does not match is never used.
- You can cancel a download and retry later. A partial download never counts as installed.
- Removing a model frees disk space and keeps all transcripts.
Using live transcription
Live Meeting is experimental. It has been tested with synthetic speech on one Mac, but not yet in real meetings.
- Open Live Meeting (⌘3). Nothing is captured when you open it.
- Check the sources: System audio and Microphone are both on by default, each with its permission status.
- Choose the spoken language and an installed model (Base is recommended), and optionally a title.
- Press Start transcription. A LIVE indicator and timer appear here and in the top bar on every screen.
- Use Pause and Resume as needed. Stop finishes outstanding audio, saves the transcript and opens it for review.
- Wear headphones. With speakers, your microphone also hears the other participants and their words may appear twice.
- Source labels mean “heard on this source”, not who spoke. Vynote does not identify speakers.
- If one source stops working, capture pauses so you can continue with the other source, retry or stop.
- Finished sessions are listed under Recent live sessions, including after you restart Vynote.
You are responsible for getting consent from everyone you transcribe and for following the laws that apply to recording conversations where you and they are.
Microphone and system-audio permissions
Live Meeting needs two macOS permissions — one for each source. File transcription needs neither.
| Source | macOS permission | How it is requested |
|---|---|---|
| Microphone | Microphone | When you choose “Allow microphone access” in Live Meeting |
| System audio | Screen & System Audio Recording | When you choose “Request permission”; you may need to enable Vynote in System Settings |
- If a permission was denied, choose Open System Settings in Live Meeting, or go to System Settings › Privacy & Security.
- Open Microphone or Screen & System Audio Recording (on macOS 14: Screen Recording) and turn on Vynote.
- If macOS asks, quit and reopen Vynote so the change takes effect.
The system-audio permission is named after screen recording because macOS uses the same framework (ScreenCaptureKit). Vynote requests audio only: no screen image is captured, shown or stored.
Understanding provisional and finalized text
Live transcription works in short overlapping windows, so the newest words are a first guess.
- In progress text is shown in a lighter style. It may still change as more audio arrives.
- Finalized text has settled. It gets a session timestamp and a source label, and it is what Vynote saves.
- When you pause or stop, Vynote finishes the speech it already captured before it settles the last lines.
- If transcription falls behind, Vynote says so. If audio had to be skipped, the gap is marked rather than hidden.
Timestamps and wording are approximate. Check important details against your own notes.
Privacy and local storage
Everything Vynote saves stays in your user Library, readable only by your macOS account:
~/Library/Application Support/work.fikrimarzuki.vynote/
media-v1.json references to imported files
transcription-v1.json transcripts, edits and settings
live-sessions/ finalized live-session text
models/ downloaded models- Original recordings are not copied there; Vynote only stores a reference.
- Live audio is never written to disk. Only finalized text and timing information are saved.
- During a file transcription, a temporary audio copy (about 230 MB per hour of media) is created and removed when the job ends.
- Data is stored as plain files, protected by your account permissions and FileVault if you use it — Vynote does not add its own encryption.
To remove all Vynote data, quit Vynote and delete the folder above.
Troubleshooting
- System audio shows “No audio received”.
- Check that Vynote is enabled under Screen & System Audio Recording, then quit and reopen Vynote. Make sure the meeting audio is actually playing on this Mac.
- The microphone does not capture.
- Check that Vynote is enabled under Privacy & Security › Microphone. Vynote uses the default input device set in System Settings › Sound.
- The same words appear twice.
- Your microphone is picking up your speakers. Use headphones, or start with only one source.
- Live text falls behind.
- Switch to the Base or Tiny model. Small is often too slow for two sources at once.
- A file cannot be imported.
- Check that it is MP4, MOV, MP3, M4A or WAV, under one hour, and has an audio track. Some codecs cannot be read by macOS.
- The transcript says “No speech detected”.
- The recording is silent or nearly silent. Check that the right audio track contains speech.
- A model download fails.
- Check your internet connection and free disk space, then try again. Interrupted downloads restart from the beginning and are verified before use.
- The recording is shown as missing.
- The original file was moved, renamed or deleted. Import it again from its new location.