Meetings

How Merriweather listens to and remembers a meeting, and how to name the speakers afterwards.

Listen and Remember

Choose Listen and Remember in the menu. The mark turns orange and the live transcript opens at the top right. Choose Stop Listening when the meeting ends.

The live transcript shows words as they are spoken. You can pause, add a note, or stop from it. Pause silences both sides. These live words are a first pass. After you stop, Merriweather reads the whole meeting again, and the final text can differ. The Transcript menu item brings the panel back if you minimized it.

Starting on its own

Settings, under Meetings, has Remember meetings automatically. It is on when the app starts. Merriweather sees a meeting when a Zoom, Microsoft Teams, FaceTime, or Google Meet window is open and the microphone is in use.

• A meeting on your calendar starts on its own. About five minutes before, a notification asks whether to listen. No answer means yes.
• Any other conversation is asked about first. If a notification cannot reach you, it starts.
• Nothing starts while tracking is paused.
• If you stop a meeting or say not now, it will not start or ask again until that meeting window closes.

If it sees no sign of the meeting for two minutes, it asks Is the meeting still going?

What is recorded

Audio is recorded in two parts: your microphone, and the sound from the Mac, which is the other side of a call. Muting your microphone, in Zoom or on the Mac, pauses your side only.

There is no video and no screenshot. Whisper turns speech into text on this Mac, and speakers are told apart on this Mac.

By default the audio is kept so a meeting can be processed again. Turn off Keep meeting audio so memories can be reprocessed in Settings to delete it once the text is written.

After the meeting

The panel shows the meeting's name and Thinking & Remembering, with a bar. The text is written in the background, so you can start the next meeting. When it is done the panel says Remembered. You can remember a new meeting, keep remembering the same one, or choose Done.

The meeting then appears in Knowledge. Its title, summary, people, and commitments are yours to edit. Corrections never rewrite the transcript.

Naming speakers

Unnamed voices show as Speaker A, Speaker B, and so on. In the meeting's speaker list, type a name over a voice and press Return. Or choose the speaker beside a line and pick from the people invited. Where the audio is kept, each line has a play button, so you can hear it first.

Naming a voice names every line it spoke. The voice bank remembers it for the next meeting in the same engagement. Rename changes it later.

Reprocessing and files

Reprocess with current engines reads the audio again with the current speech and speaker models. Show files opens the meeting folder.

Each meeting has a folder under Meetings, by engagement and month. It holds mic.wav, system.wav, transcript.json, and metadata.json. The engagement's voice bank, voices.json, sits beside the month folders.

An interrupted meeting is read again in the background. If the text cannot be written, the audio is kept for another try.