Recognize people in later meetings
Saved-voice recognition can put a known person's name into later meetings. It is off by default. Ordinary speaker separation and naming still work without it; recognition adds a voice-derived record kept for future matching.
Turn on recognition
Section titled “Turn on recognition”- Open Settings (⌘,) › Speakers › Saved voices. When recognition is off, the speaker review also offers Open Settings and points to this switch.
- Turn on Recognize people across meetings.
- Read the explanation, check I will ask the people whose voices this saves., then click Turn On.
Opening Settings or turning on this switch saves no new voice by itself. The first time you name a new person in a speaker review, Save these voices? asks which of those people may be kept. Check only those you can attest for, then click Save. An unchecked person is still named in that meeting, but no voice record is saved for later recognition. Editing an old meeting's name without new voice evidence also does not retrain a saved voice. If that edit rejects a voice confirmation from the meeting, Recordstead stops using that confirmation for future matches and asks for new evidence before saving the corrected voice.
Your result: eligible saved voices can be matched in later meetings in the same workspace. A meeting where every voice is confidently recognized can finish without asking you to name speakers; you can edit the names afterwards.
Turn the setting off to stop using saved voices for matching. Existing records remain until they expire after 90 days of inactivity. Deleting them sooner removes every saved voice in the current workspace and cannot be undone; the names in your meetings stay. To do it, click Delete Saved Voices in People, then Delete. What leaves your Mac explains where this data is kept.