How speakers get their names

After a meeting, embral separates the voices in the recording and labels them Speaker 1, Speaker 2, and so on, in order of first appearance. With the accurate local model or cloud transcription, provisional labels appear live during the meeting too.

embral doesn’t recognize voices. There’s no stored voiceprint and no matching of a voice heard today against one heard last week. That’s deliberate: it keeps the app from guessing identities, and it means a person is only ever named because you named them.

Where names come from

Two places:

• You. Rename a speaker label in any transcript, either live or afterward. See Name speakers in a transcript.
• Your notes. If you typed something like Ana: we need a new plan, embral matches that line against the transcript and suggests the name for the matching speaker.

How a name carries across meetings

Naming a speaker creates a profile: a name plus any notes you add. When labels in other meetings match that name, they link to the same profile, so the Profiles page can show every meeting a person spoke in and what they said. Generic labels like Speaker 2 never link to anything.

If the same person ends up with two profiles (“Ana” and “Ana Silva”), merge them on the Profiles page. Every meeting moves to the profile you keep.