Skip to main content

People & voiceprints

Draft — page in progress

This feature is documented briefly here and will be expanded. Have feedback? support@scribx.app.

Screenshot placeholder
📷 To capture: a meeting's People tab showing participants and per-speaker labels, with a speaker being renamed (target: /img/handbook/web-app/people.png).

Every meeting has a People tab showing who took part and how their speech was labeled throughout the transcript. It's the place to check attribution and fix any names before you share your notes.

Speaker identification

ScribX automatically separates speakers — a process called diarization — and labels each transcript segment with the person who said it. This is what lets the transcript, summary and other notes attribute quotes, decisions and action items to the right people.

Attribution isn't always perfect: overlapping speech or brief interjections can be mislabeled. The People tab lets you correct this.

  • Rename a speaker and the change applies everywhere — across the transcript and the notes generated from it.
tip

Renaming speakers before you share is the quickest way to make notes readable for people who weren't in the room.

Voiceprints Enterprise

Optionally, ScribX can learn a person's voiceprint — a model of how their voice sounds — so they're recognized by name across future meetings automatically, without renaming each time.

Voiceprints are treated as biometric data, and ScribX handles them accordingly:

  • Consent required. Enrolment always requires the person's consent.
  • Admin-controlled. The scope of voiceprints and whether auto-enrolment is enabled are configured in org settings.
  • Never shared, never used for training. Voiceprint data is not shared and is never used to train models.
note

Because voiceprints are biometric identifiers, they carry specific legal obligations in some regions. See Data & retention and our DPA for our BIPA/CUBI commitments.

What's next?