Disable speaker enrollment
When this setting is enabled for your workspace, Otter will disable automatic speaker enrollment using stored voice data to identify or label speakers in recordings. Speaker labels will instead rely on participant names provided by your video conferencing platform or on manually assigned speaker tags.
This setting is designed for organizations with privacy, security, or compliance requirements related to voice data and speaker recognition.
What changes when this setting is enabled
Otter will not:
- Create or store speaker enrollments for workspace members
- Use existing voice data to automatically identify or label speakers
- Automatically re-label speakers using stored voice data after a meeting ends
Otter will still:
- Separate transcripts by speaker turns through speaker diarization (e.g., unknown or Speaker 1, Speaker 2, Speaker #)
- Label speakers using participant names provided by your video conferencing platform, where available
- Allow users to manually tag and label speakers in transcripts
- Preserve any manual labels you apply — these are never overwritten
What this setting does not change
Speaker segmentation still runs. Otter still separates the transcript into turns and groups bubbles by speaker — it just won't use stored voice data to identify who those speakers are.
Meetings recorded by other organizations. If a workspace member joins a meeting recorded by another company's Otter workspace, that workspace's settings govern the recording. Learn more about managing your workspace’s auto-join settings.
Data retention is separate. This setting disables the creation and use of speaker enrollments going forward. If you also need to delete existing enrollment data, that is managed through your workspace's data retention settings. Learn more about setting up a custom data retention policy.
How speaker labeling works during a meeting
| Recording type | During a live recording | Speaker labels after conversation processing |
|---|---|---|
| Otter Notetaker meeting (Zoom, Google Meet, Microsoft Teams, etc.) | Participant names may pull from your video platform associated with the profile. | Participant names from your video platform appear where they can be matched to a speaker turn. Unmatched turns show as unknown, or Speaker 1, Speaker 2, etc. |
| Ad-hoc or uploaded recording | No speaker enrollment or automatic tagging. Speakers show as unknown. | No participant data is available, so all speakers show as unknown, or Speaker 1, Speaker 2, etc. |
These names come from your video platform's own participant list — not from Otter's voice recognition. No speaker enrollment is created or used in this process.
Notetaker (Zoom, Google Meet, Microsoft Teams, etc.)
During a live meeting, Otter attempts to label speakers in real time using participant names provided by your video conferencing platform (e.g., Zoom, Google Meet, or Microsoft Teams, and others). As participant information becomes available, speaker names may appear on speaker bubbles during the conversation.
Because automatic speaker enrollment using stored voice data is disabled, Otter cannot verify those live speaker labels from the video platform using voice matching. As a result, some speakers may temporarily appear as unknown until the meeting ends and final transcript processing is completed.
These names come from your video platform's own participant list — not from Otter's voice recognition. No speaker enrollment is created or used in this process.
Manual & uploaded recordings
For ad-hoc manual recordings or uploaded files, Otter will not automatically label speakers during the recording or after conversation processing. Speakers will continue to appear in the conversation as Unknown or as generic speaker labels (Speaker 1, Speaker 2, etc.).
Manually labeling speakers Learn more about tagging a speaker
You can replace any Speaker # or platform-provided label with a name at any time. You can apply the label to a single bubble or to all bubbles with the same speaker grouping in that transcript. For example,
- Conversation contains multiple Speaker 1 bubbles
- Tagging Speaker 1 as Dani will update all Speaker 1 bubbles in the conversation as Dani
Manually labeling a speaker does not create a speaker enrollment. The label is stored as a text tag only — it will not be used to identify that person in other meetings and will not carry over to future transcripts.
Speaker Rematch Learn more about speaker rematching
Speaker Rematch re-processes speaker labels across a transcript after you've made manual edits. With this setting enabled:
- Your manual labels are applied and preserved
- No stored voice data is used at any point in the process
- Your manual labels will not be overwritten
FAQs & Info
Speaker glossary terms
You may see these terms in Otter or hear them referenced by your Otter account manager. Refer to the glossary below to learn about each term.
Speaker Enrollment: Refers to the stored voice data used by Otter to help automatically identify and label speakers across conversations. Disabling speaker enrollment will stop automatic tagging based on existing voice data.
Speaker Diarization: The process of separating a transcript into different speaker turns (speaker profile bubbles), such as Speaker 1, Speaker 2, etc.
Speaker Turn: A speaker turn reflects the sections of a transcript associated with a specific speaker label. In Otter, each speaker's turn appears as a speaker profile bubble paired with the portion of the transcript attributed to that speaker.
Will speakers be identified automatically?
Partially. Otter will use participant names from your video conferencing platform to match them to speaker turns. Any turns that can't be matched will appear as "Speaker 1", "Speaker 2", etc. Otter will not attempt to cross-reference past recordings.
Do the manual labels I apply in one meeting carry over to future meetings?
No. With this setting on, labels are scoped to the individual transcript and are not used to recognize that person in future recordings.
What if my workspace had existing speaker enrollments before this setting was turned on?
Existing enrollments are ignored — they are not used to label speakers in any new recordings. They are not deleted unless you separately configure a data retention policy to remove them.
Does this affect Action Items, summaries, or Otter AI Chat?
No. Speaker enrollment is only used for speaker labeling in transcripts. All other AI features are unaffected.
Feedback
0 comments
Article is closed for comments.