Help centre › Team, roles & security
Voice ID: speaker recognition and voice security
How your colleague tells voices apart, names speakers in meetings and calls, and protects your personal information.
Updated 14 October 2026
Voice ID lets your digital colleague recognise who is speaking.
What it does
- Meeting notes and calls: transcripts are labelled by speaker. When people introduce themselves or are named, the agent uses their names; otherwise you see Speaker 1, Speaker 2 and so on. Actions are assigned to the right person.
- Bring someone in: in a live voice chat, say "I have Mark with me" and the agent talks to Mark as Mark when he speaks.
- Security lock: if the agent hears a voice that is not yours, it will not read out your emails, contacts or anything personal until you are speaking again.
Saving voices (consent)
A voice is only saved to your private voice vault after the person clearly says yes when the agent asks "Are you happy for me to store your voice?". If they say no, nothing is kept.
A saved voice is a voice print (a mathematical pattern), not a recording. It is processed on our own UK servers, never shared or sold, and never used to train AI.
Manage your voices
Go to My profile → Voices to see saved voices, switch Voice ID on or off, and delete any voice at any time. Voices are deleted when your account closes.
Availability
Voice ID is switched on for everyone by default. To turn it off, go to My profile → Voice ID → Manage and switch it off. You can reset your own voice at any time from your profile.