Correcting Speaker Names in Transcripts

Groupthink separates a meeting's audio into individual speakers as it transcribes. Until you tell it who's who, those speakers usually show up as generic labels like Mic Speaker 1 or Conference Speaker 1. On a simple one-on-one, Groupthink may fill in the name for you and mark it Suggested (see Suggested names on one-on-ones). Name a speaker once and every line that speaker said, earlier in the meeting and for the rest of it, switches to that person's name and avatar.

This matters beyond a tidier transcript. The name you assign is the attribution Groupthink uses to build each person's profile and surface what they said in your briefs. Getting it right means the right intel lands on the right person.

Name a speaker

You can do this during a live meeting or afterward on the recap.

1. Open the transcript. During a live meeting the transcript is on screen as people talk. On a finished meeting, open the Recap and go to the Transcript.
2. Click a speaker label, for example Mic Speaker 1. A menu opens, headed Who is "Mic Speaker 1"? (or Reassign "Mic Speaker 1" if the label already has a name). If the name has a Suggested chip, the menu instead tells you which person Groupthink assumed and offers Yes, this is [Name] above the list.
3. Pick the person from the list. When the list is long, type in the Search people field at the top to narrow it.

Every line from that speaker updates to that person's name and avatar right away. There's no separate save step.

Who can name a speaker

You can name a speaker if you were in the meeting, or if you are a Host or Participant on its agenda.

If you can open the meeting but are neither, for example you have the link to it or you are a Viewer who was not in it, you see the speaker names without the menu. Clicking a label does nothing. To name speakers, ask the meeting's host to add you to the agenda as a Participant. See Managing Agenda Permissions.

Who shows up in the list

The list is not limited to people who were on the call. You can pick from:

• People on the meeting: Groupthink users who attended, then people linked to the meeting.
• Your saved contacts, so you can attribute a voice to someone in your relationships even if they don't have a Groupthink account or weren't formally on the invite.

Naming a speaker as one of your contacts stays private to you. The person is not notified, and other people on the meeting don't see your contact or the name you use for them. They see the name on that person's Groupthink account if they attended, or the name on the calendar invite. If the meeting has neither, they keep seeing the generic label.

Anyone who can name speakers on the meeting can rename a speaker you named, and their choice replaces yours.

Suggested names on one-on-ones

When a meeting ends, and again when you link a person to a meeting that has ended, Groupthink checks whether it can safely name the other person for you. It does this only when the call had exactly one unnamed Conference Speaker and exactly one of the host's contacts linked to the meeting besides the host. In that case it names that speaker as the linked person and adds a Suggested chip next to the name, so you can tell it was assumed rather than chosen by you.

• To accept it, click the name and choose Yes, this is [Name]. The chip goes away.
• To fix it, click the name and pick someone else, the same way you would name any speaker.

If the call has more than one unnamed Conference Speaker label, if more than one of the host's contacts is linked to the meeting, or if that contact already names another speaker, Groupthink does not guess. The labels stay generic until you name them.

Names Groupthink remembers

When a speaker shows up under a real name, for example Karlie Peterson, and you name that speaker as one of your contacts, Groupthink remembers it. The next time a meeting you were in ends with a speaker called Karlie Peterson, that speaker is already named as your contact, with the Suggested chip. Confirm it with Yes, this is [Name], or pick someone else.

• Only real names are remembered. Generic labels like Mic Speaker 1 or Conference Speaker 2 are a different person in every meeting, so naming one never carries over.
• Changing your mind updates it. Name the speaker as a different contact and that contact is suggested from then on. Name them as someone with a Groupthink account and the contact stops being suggested.
• It only fills in names nobody has set. A speaker already named on that transcript, or matched to an attendee's Groupthink account, is left as it is. If two people on the meeting have each linked the name to a different contact, Groupthink does not guess.
• It starts from now. Names you set before this was added are not remembered until you set them again.
• It stays private. Like any contact you name a speaker as, other people on the meeting don't see your contact.

Good to know

• It applies to the whole meeting. Naming a label reassigns all of that speaker's lines at once, not just the one you clicked.
• It keeps working for the rest of the meeting. New lines from a named speaker come in already attributed.
• You can change it. Reassign a label to a different person and the lines move over.
• It syncs. If a teammate has the same meeting open, their transcript updates too, without a reload. When you picked one of your own contacts, they see the meeting's name for that speaker, not yours.
• Contact names are private. Attributing a label to one of your contacts is yours alone, and they aren't notified. On a meeting with other Groupthink users, the recap everyone reads is written with the meeting's names for people, not any one person's contact names.

Why a speaker stays unlabeled

A label like "Mic Speaker 1" just means Groupthink heard a distinct voice it couldn't match to a name on its own, usually someone dialing in on a shared line or a device whose audio doesn't carry a participant name. Name it yourself and it's fixed for good.

Overlapping speech, heavy background noise, or a room mic picking up several people can also blur who's speaking. A headset or dedicated microphone gives the cleanest separation. See Known Issues for more on transcription accuracy.