Record a meeting
A meeting recording captures a whole conversation, not just your voice. InkSpoke records your microphone and the audio coming out of your computer's speakers at the same time, then turns the two into a single speaker-labeled transcript — all on your own device. It's the tool for calls, interviews, and any long stretch of talking that push-to-talk dictation isn't built for.
You start and stop a meeting from the system tray, and a small coral meeting HUD walks you through three phases: get ready, record, and process. This page is a complete tour of that flow.
Starting and stopping from the tray
There's no global hotkey for meetings — everything runs from the tray menu, so it's out of the way until you want it.
┌─────────────────────────────────┐
│ InkSpoke │
│ ───────────────────────────── │
│ ▶ Start Meeting Recording │ ← flips to "Stop Meeting Recording" while active
│ 📄 Transcribe a File… │
│ 🕘 History… │
│ … │
└─────────────────────────────────┘
- Click the InkSpoke tray icon and choose Start Meeting Recording. The meeting HUD opens in its Ready phase.
- Pick your options (below) and click Start Meeting to begin capturing.
- When you're done, click Stop on the HUD — or open the tray menu again, where the item now reads Stop Meeting Recording.
The tray toggle stays on Stop Meeting Recording for as long as the meeting is recording or processing, so you always have a way to end it.
On macOS and Linux these are native menu items; on Windows they live in InkSpoke's tray popup. The labels and behavior are identical everywhere.
The meeting HUD, phase by phase
The HUD is a frameless, always-on-top coral window titled Meeting Recording. Drag it by its title strip to reposition it; the ✕ closes it. It moves through three phases.
1. Ready — choose how this meeting is captured
Before anything records, you set the options for this session. None of these are saved as defaults — they only apply to the meeting you're about to start.
┌────────────────────────────────────────────────┐
│ ● Meeting Recording ✕ │
├────────────────────────────────────────────────┤
│ │
│ [ Workspace ▾ ] [ Language: Auto ▾ ] │
│ [ Model: Whisper Small (local) ▾ ] │
│ [✓] Identify speakers (diarization) │
│ │
│ [ Cancel ] [ Start Meeting ] │
└────────────────────────────────────────────────┘
| Picker | Default | What it does |
|---|---|---|
| Workspace | Your default | Applies a workspace's vocabulary and dictionary substitutions to the transcript. |
| Language | Your preferred language, or Auto | Sets the spoken language (or lets InkSpoke detect it). |
| Transcription model | Your active local Whisper model | The on-device model used to transcribe. Only local models are offered here. |
| Identify speakers (diarization) | On | Labels who spoke when. Turn it off to keep the whole recording as one speaker. |
The model picker in the meeting HUD lists local models exclusively. Live meetings run fully on your device — cloud (Platform) transcription for meetings is coming soon. If you need a cloud model, use Transcribe a file instead, where cloud transcription is available today. See On-device vs. cloud for why on-device is the default.
2. Recording — watch it capture both sides
Click Start Meeting and the HUD switches to a live view: an elapsed timer, two level meters, and a rolling preview of what's being said.
┌──────────────────────────────────────────────┐
│ ● Meeting Recording ✕ │ ← drag this strip to move
├──────────────────────────────────────────────┤
│ 2:47 │ ← elapsed timer (m:ss)
│ │
│ You (mic) ▇▇▇▅▃▁ │ ← your microphone level
│ Participants ▇▇▇▇▇▆▄▂ │ ← system-audio level
│ │
│ "…so the rollout is scheduled for Friday │ ← rolling ~4s live preview
│ and the docs still need a review pass." │
│ │
│ [ Cancel ] [ Stop ] │
└─────────────────────────────────────────────────┘
- The timer counts up in
m:ss. - You (mic) and Participants (system) are two separate level meters — one for your microphone, one for the audio your computer is playing (the people on the other end). Seeing both move confirms you're capturing the whole conversation.
- The rolling preview shows a running snippet of the transcript so you know it's hearing everyone. When there's nothing to show yet, it reads an italic Listening….
- Stop ends the meeting and moves to processing. Cancel throws the session away without transcribing.
The rolling text is a quick, non-authoritative peek — it can drop or garble words. The clean, speaker-labeled transcript is produced after you press Stop, from the full recording. Don't judge accuracy by the preview.
Every ~4 seconds InkSpoke transcribes the trailing ~7.5 seconds of mixed mic + system audio using a separate engine from your dictation, purely for the on-screen preview. It filters out likely hallucinations; if a whole window is filtered away, the preview resets to Listening…. Because both tails are mixed, remote participants who only come through your speakers still show up in the preview.
3. Processing — building the transcript
After Stop, the HUD shows a short processing state while InkSpoke transcribes the full recording and separates the speakers:
┌────────────────────────────────────────────────┐
│ ● Meeting Recording │
├────────────────────────────────────────────────┤
│ Transcribing… │
│ │
│ Diarizing speakers and building the │
│ transcript. │
└────────────────────────────────────────────────┘
When it finishes, the Transcript Viewer opens automatically with your speaker-labeled transcript, ready to rename speakers, copy, or export.
Capturing system audio (the other participants)
The magic of a meeting recording is that Participants (system) track — it grabs whatever your computer is playing so the people on your call are transcribed too. How that capture works depends on your OS:
| Platform | How system audio is captured | Setup |
|---|---|---|
| macOS | ScreenCaptureKit | Requires Screen Recording permission |
| Windows | WASAPI loopback | Works out of the box |
| Linux | PulseAudio / PipeWire monitor | Works out of the box |
On macOS, capturing system audio goes through ScreenCaptureKit, which requires Screen Recording permission. Grant it under Settings → Recordings & Meetings → System Audio → Grant Screen Recording… (or in macOS System Settings). Until you do, meetings record your microphone only.
If system audio isn't available — permission denied on macOS, or an unsupported device or build — the meeting doesn't fail. It records your microphone only, and the HUD shows a warning banner so you know participants aren't being captured:
│ ⚠ System audio unavailable — recording your microphone only │
You can check availability ahead of time under Settings → Recordings & Meetings → System Audio, which shows a green Available, a Needs permission prompt (macOS), or Not available on this device — meetings will record the microphone only.
Keeping the audio and pruning old recordings
By default InkSpoke keeps only the transcript, not the audio. If you want to replay a meeting later, turn on Keep recorded audio before you record. These live in Settings → Recordings & Meetings:
| Setting | Default | What it does |
|---|---|---|
Keep recorded audio (Recording.KeepAudio) | Off | Saves the mixed meeting audio alongside the transcript so it can be replayed. The audio is written before transcription, so even if transcription fails the recording is safe. |
Keep recordings for (Recording.RecordingRetention) | Keep forever | Auto-prunes recordings older than the chosen period (moving audio to the trash). Options: Keep forever, 1 week, 1 month, 3 months, 6 months, 1 year. |
Identify speakers (Recording.DiarizationEnabled) | On | The global default for the diarization checkbox in the HUD. |
Processing (Recording.ProcessingMode) | On device (Local) | Where meeting transcription runs. Cloud (Platform) is shown as coming soon for meetings. |
The Transcript Viewer's Play/Stop button only appears when there's stored audio. With Keep recorded audio off, you'll still get a full transcript — you just won't be able to replay the recording. Flip it on before recording anything you might want to hear again.
Telling participants apart
Diarization is what turns a wall of text into a labeled conversation. In a meeting, the two capture tracks already do a lot of the work: your microphone becomes You and the system audio becomes Participants. To split those participants into individual speakers, download InkSpoke's on-device speaker model — a one-time ~46.5 MB download under Settings → Recordings & Meetings → Speakers → Download diarization model. Speaker identification then runs entirely on your device.
Meetings auto-detect how many people are on the call — there's no manual speaker-count picker in the meeting HUD. (That picker lives in the file-import dialog.) For a full tour of renaming speakers and reading the transcript, see Speakers and the transcript viewer.
Next steps
- Transcribe a file — already have a recording? Import audio or video, with cloud transcription available.
- Speakers and the transcript viewer — rename speakers, replay audio, and export to TXT, Markdown, SRT, VTT, or JSON.
- History and diagnostics — find past meetings, filter by type, and reopen any transcript.
- Tutorial: Transcribe a meeting — a start-to-finish walkthrough.