Diarization separates speakers
Diarization groups speech by speaker within recorded audio. It answers which voice spoke when. A label such as Speaker 2 does not, by itself, establish a person’s name.
Recognition proposes an identity
Speaker recognition compares available voice information with known speaker profiles. A proposed link can connect an utterance with the person’s history and reminders. Similar voices, short samples, background noise, and overlapping speech can reduce reliability.
Review uncertain assignments
The app groups uncertain segments for review. Check their sample text and surrounding context before confirming or correcting the identity. A mistaken assignment can make a correct sentence appear in the wrong person’s history.
Build useful references
Use representative, clean speech when reviewing or training a speaker profile. A name mentioned in a sentence is not necessarily the person speaking. Confirm the relationship rather than relying solely on name guessing.
Identity and command permission differ
Remembering a voice is not a universal authorization guarantee. Speech activation, command gating, wearer enrollment, and tool permissions are separate mechanisms whose availability depends on the route and configuration. Always check the confirmation required for the intended action.
Try it with a sample
- Open a saved conversation with more than one speaker.
- Inspect uncertain speaker assignments.
- Confirm the correct profile, then recheck the person’s history.
About this explanation
Reviewed against Modules/Assistant/Shared/Services/SpeakerReviewService.swift. Examples use fictional people and content. Availability depends on the build, edition, permissions, and selected services. Source changes trigger a review notice and removal from website AI retrieval until reviewed.