PulseSpeaker labels
How speech is attributed to each speaker, what can and cannot be identified, and how to think about telling participants.
Speaker labels show the name of the participant who said something above each passage of the transcript. By default speech is only split into “me” and “them”, but with speaker labels set up, the display name from the meeting tool is attached as the name. This helps when reading back the record of a meeting with many participants, or checking whose task something was agreed to be. This article covers how it works, which meeting tools are supported, the limits of identification, and how the recorded names are handled.
Participant names shown against each block of speech.
Speaker labels are a desktop app (macOS / Windows) feature and work only while transcription is running. Names are never added retrospectively to a transcript recorded before you set it up. The setup required differs by meeting tool.
How speaker labels work
Speaker labels read the participant names shown on the meeting tool's screen, along with the indicator showing who is currently speaking, and match them against each section of the transcript. There is no processing that infers who someone is from voiceprints or audio characteristics. What the meeting tool displays therefore determines the result directly.
How that reading is done differs by meeting tool: Google Meet uses a browser extension, Zoom uses macOS accessibility permissions. In neither case do speaker labels record any new audio or video. Audio capture for transcription happens within the scope described in How transcription works.
Supported meeting tools and how to set them up
| Meeting tool | Supported OS | What you need | Setup steps |
|---|---|---|---|
| Google Meet (Chrome or a Chromium-based browser) | macOS / Windows | the Pulse browser extension | Using speaker labels in Google Meet |
| Zoom (the Zoom Workplace desktop app) | macOS only | macOS accessibility permission | Using speaker labels in Zoom |
| Microsoft Teams, Zoom in a browser (web client) | — | not supported at this time | the transcript shows “me” and “them” |
To deploy across company machines, Setting up speaker labels across your organisation (for admins) covers distributing the browser extension in bulk and pre-granting permissions through MDM.
What to do during the meeting
Once set up, nothing special is needed during the meeting. Join as usual and start transcription, and participant names appear in the transcript panel.
You do still have to start transcription separately from speaker labels. Pulse transcribes only when you open a note from a notification or an event, or create a new note. Setting up speaker labels does not by itself cause recording to start as you join a meeting. The conditions for starting are covered in How transcription works.
What can and cannot be identified
| Situation | How speaker labels behave |
|---|---|
| Each participant joins individually from their own machine | The name shown in the meeting tool is used as-is. |
| Several people join from one machine in a meeting room | Several people speaking through the same machine cannot be told apart individually. |
| Audio other than the meeting is playing on the computer | It may appear in the transcript, but no speaker name is attached. |
| Several people speak at once | Some speech may not be attributed to the right speaker. |
| A participant uses a display name that is not their real name | That display name becomes the speaker name. |
Where speaker labels are unavailable or not set up, the transcript is split into “me” and “them”. This corresponds to the distinction between audio from your microphone and system audio (what your computer is playing). Even then, notes are often produced by inferring who spoke from the flow of the conversation and how people address each other; if a name is wrong you can rewrite it directly in the note or tell chat to correct it.
Where a speaker name is wrong you can edit that part of the note directly. If the same error runs through the whole transcript, telling chat to “replace the speaker name X with Y” fixes them all at once. On recognition accuracy for technical terms and company names, see also Customising transcription.
Speaker labels and personal information
The name recorded as a speaker label is the participant's display name in the meeting tool. Where a real name is used as the display name, the speaker name left in the transcript may constitute personal information linked to what was said. Under the Act on the Protection of Personal Information, where personal information is acquired through recording or transcription, the purpose of use must be promptly notified to the individual or made public, unless it has been published in advance. There is no general provision in Japanese law requiring everyone's consent for a participant to record a meeting, but given that the record also captures who said what, telling people at the start of the meeting what the recording is for and how long it is kept makes problems less likely later. The reasoning and sample wording are covered in How far you can rely on AI meeting notes (a guide for teams).
Letting participants know that Pulse is in use is the job of the features described in About the transparency features, not of speaker labels. If you want a notice shown during the meeting, set that up instead.
Turning speaker labels off
If you do not want speaker labels, remove the setting used to read them. In Google Meet, delete or disable the browser extension; in Zoom, switch off the macOS accessibility permission. The transcript then goes back to showing “me” and “them”. For how to exclude particular departments across an organisation, see Setting up speaker labels across your organisation (for admins).
If something is not working
Q. I set it up but no speaker names appear
A. First check the meeting was held in a supported meeting tool: Google Meet requires a supported browser, Zoom requires the macOS desktop app. Then check that the relevant row under Settings > Preferences > Speaker labels is set to enabled. Detailed steps are in Using speaker labels in Google Meet and Using speaker labels in Zoom.
Q. Can I add names to a past transcript?
A. No. Speaker labels work only while recording, so they do not apply to transcripts created before you set them up. For past notes, add the speaker names in the note yourself or ask chat to tidy it up.
Q. Only some passages have names
A. This happens where the meeting tool was not clearly indicating who was speaking at that moment, or where several people were speaking from a shared machine in a meeting room. Passages containing audio from outside the meeting also get no name.