Speakers and voice profiles
Every recording is split by speaker automatically. Names come from the meeting, from your team's voice profiles, or from you, and you can always correct them.
- Updated
- 8 October 2026
- Read
- 4 min
- For
- Everyone; profiles need a paid plan
Four ways a speaker gets a name.
Separation is automatic; naming depends on what Citesvue already knows.
- From the meeting
- When the assistant captures a call, speakers are matched to the meeting participants, so the transcript says who spoke by name without you doing anything.
- From your voice profiles
- Once someone has a voice profile with enough samples, Citesvue recognises them in later recordings. Confident matches are named for you; less certain ones show Verify identity, so a person confirms them.
- By tagging
- In the speakers panel on a recording, play a sample and choose Tag speaker. Pick an existing profile, create a new one, or mark the speaker anonymous.
- Not yet
- A speaker nobody has named or matched shows as "Speaker 1", "Speaker 2", and so on. That is honest rather than a guess, and tagging fixes it in seconds.
How profiles recognise your team.
A directory of people
A voice profile is a person in your workspace: a name, and optionally an email and role. Settings, then Voice profiles, lists them with their samples, match quality, and when they were last heard. Included on paid plans.
Better with every sample
Each confirmed tag adds a sample to the profile. Match quality reads Needs samples, Good, or Excellent, and a profile becomes matchable after you identify the person in a processed recording.
Kept in your workspace
Profiles and their voiceprint data stay in the workspace that created them. Removing a profile unlinks it from recorded speakers without removing any recording.
People check the uncertain ones
Only high-confidence matches are applied automatically. Lower-confidence matches stay visible for someone to verify, and unmatched speakers stay separate until a profile has enough samples.
Correct speaker names improve answers, because a citation can then say who said something. See ask your evidence.
Speakers, answered.
Only through the voice profiles your workspace creates. When you tag a speaker, the profile keeps a voiceprint from the confirmed samples and uses it to recognise the same person in later recordings in this workspace. Confident matches are named automatically; less certain ones wait for someone to verify them.
It names the speaker on this recording and adds a sample to their profile, so later recordings can recognise the same voice. A guest you tag once becomes a profile too; mark a speaker anonymous if you would rather not name them.
Per-speaker labels are still rolling out for Webex. The transcript itself is complete, and you can tag speakers by hand.
It depends on plan: the free plan has none, and paid plans include a directory that grows with the tier. The pricing page lists the number for each plan.
Yes, from Settings under Voice profiles. Removing a profile unlinks it from the speakers it named; the recordings themselves are not removed.