> ## Documentation Index
> Fetch the complete documentation index at: https://docs.buildbetter.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Speaker Assignment

> Assign speakers to identify who said what in your recordings

Speaker assignment helps you identify who said what in your recordings. BuildBetter attempts automatic speaker detection, but you can manually assign speakers when needed for accuracy.

## How Speaker Assignment Works

<CardGroup cols={2}>
  <Card title="Automatic Detection" icon="robot">
    BuildBetter attempts to match speakers using meeting participant data from calendars and platforms
  </Card>

  <Card title="Manual Assignment" icon="user">
    You can manually assign unknown speakers to people in your workspace
  </Card>
</CardGroup>

## Unknown Speakers Tab

When a recording has speakers that couldn't be automatically identified, an **Unknown Speakers** tab appears in the call details view.

### Assigning Unknown Speakers

<Steps>
  <Step title="Open Unknown Speakers" stepNumber={1}>
    Click the "Unknown Speakers" tab in the call details page
  </Step>

  <Step title="Review Speaker Information" stepNumber={2}>
    For each unknown speaker, you'll see:

    * Screen name (username from Zoom, Teams, etc.)
    * Speaking time and segments
    * Audio samples to listen to
  </Step>

  <Step title="Assign to Person" stepNumber={3}>
    * Click "Assign Speaker"
    * Search for the person by name or email
    * Select the matching person
    * Confirm assignment
  </Step>

  <Step title="Verify Assignment" stepNumber={4}>
    * Review transcript to ensure assignment is correct
    * Speaker labels update throughout the transcript
    * Signals re-attribute to correct person
  </Step>
</Steps>

<Tip>
  Listen to audio clips of the unknown speaker to help identify them, especially when screen names aren't clear (e.g., "iPhone" or "Guest").
</Tip>

## Automatic Speaker Detection

BuildBetter uses meeting metadata to automatically assign speakers when possible:

### Detection Sources

* **Calendar participants**: Email addresses from meeting invites
* **Platform usernames**: Screen names from Zoom, Teams, Google Meet
* **Company associations**: Matching email domains to organizations
* **Previous recordings**: Historical speaker data from past calls

### When Automatic Detection Works Best

* Meeting participants are in your calendar invite
* Participants use their real names as screen names
* Email addresses match people in your workspace
* Same participants in recurring meetings

### When Manual Assignment is Needed

* Participants join with generic names ("iPhone", "Guest")
* External participants not in your system
* Phone dial-in participants
* Screen names don't match email addresses

<Note>
  The Unknown Speakers tab only appears when there are unassigned speakers. Once all speakers are assigned, the tab disappears.
</Note>

## Speaker Matching

### Person Linking

When you assign a speaker, BuildBetter:

* Links the speaker to a person record in your workspace
* Updates all transcript segments for that speaker
* Re-attributes any extracted signals to the correct person
* Associates the person with their company (if known)

### Company Association

* People are automatically linked to companies based on email domain
* You can manually set company associations
* Company data syncs from CRM integrations (Salesforce, HubSpot)

## Best Practices

<Check>
  **Assign speakers early**: Speaker assignment affects signal attribution and insights
</Check>

<Check>
  **Use real names in meetings**: Encourage participants to use their actual names as screen names
</Check>

<Check>
  **Keep people records updated**: Ensure email addresses and names are current in your workspace
</Check>

<Check>
  **Review after recording**: Check Unknown Speakers tab soon after calls process
</Check>

<Check>
  **Listen to samples**: Use audio clips to confidently identify unclear speakers
</Check>

## Troubleshooting

<AccordionGroup>
  <Accordion title="Speaker Not Found in Search">
    * Check if the person exists in your workspace (People section)
    * Verify email spelling is correct
    * Try searching by partial name
    * Create new person record if they're not in system
  </Accordion>

  <Accordion title="Wrong Speaker Auto-Assigned">
    * Open Unknown Speakers tab
    * Reassign to correct person
    * Speaker labels update automatically
    * Previous incorrect assignments are overwritten
  </Accordion>

  <Accordion title="Multiple Unknown Speakers">
    * Assign one at a time using audio samples
    * Use speaking time as a hint (who spoke most)
    * Check screen names for clues
    * Review calendar participants for guidance
  </Accordion>
</AccordionGroup>

## Privacy & Permissions

### Speaker Data

* Speaker assignments are private to your workspace
* External participants can be added as people without full access
* Speaker information respects recording privacy settings

### Access Control

* Only call participants and workspace admins can assign speakers
* Speaker assignments are permanent once saved
* Changes apply immediately to transcript and signals

<Info>
  Accurate speaker assignment improves signal attribution, enables better filtering by person, and provides clearer context for extracted insights.
</Info>