This is a feature where AI automatically completes speaker separation after you manually separate speakers to some extent.
If you have subscribed to an enterprise plan, you can use automatic speaker separation using voice data (voice prints). Please see below for details.
Flow of Using "Automatic Speaker Separation Feature (Without Registering Voice Data)"
Step1| Creating and Registering Participants
Step2| Manually Add Line Breaks for Each Speaker and Set Speaker
Step3| Execute Automatic Speaker Separation
Step1| Creating and Registering Participants
1. Click the three-line icon displayed at the top right of the page to display the side panel. Also, click the "Edit" button on the transcription to enter edit mode.
2. Click "Set Speaker" at the beginning of the paragraph and select "Add New Speaker" from the displayed options. Alternatively, click the "Edit" button on the "Speaker Separation" screen in the right side panel.
3. A popup to register participants will be displayed. Enter the participant's name and click the "Enter" key or the "Add ○○" button to register the participant.
4. After registering all participants, click "Save".
Step2| Manually Add Line Breaks for Each Speaker and Set Speaker
1. To help the AI learn the audio and speaker names, manually add line breaks to separate each speaker in the transcription results.
*Please manually separate speakers enough that each participant has at least 10 seconds of speaking time.
2. Assign registered speakers to each text section separated by line breaks.
Click "Set Speaker" displayed to the right of the timestamp in the transcription text, and assign speakers.
As you proceed with manual speaker settings, a green gauge in the "Speaker Separation" screen on the right side panel will show how many seconds have been set for each participant's speaking sections.
Step3| Execute Automatic Speaker Separation
1. Once all participants have been assigned at least 10 seconds of speaking time, the "Execute" button will become active. Click it to perform automatic speaker separation. When speaker separation is completed automatically, a message "Speaker separation has been executed" will be displayed in the bottom right.
For participants with less than 10 seconds of speaking time:
Click "・・・" to the right of the participant's name on the "Speaker Separation" screen in the side panel, and select "No more speaking sections found".
2. If speakers are not properly separated, manually add more speaking sections and click "Re-execute".
Screen After Automatic Speaker Separation
Participants: Participant names are displayed according to the speaker separation settings.
Green checkmark: Speaker setting is confirmed.
No checkmark: Speaker setting is unconfirmed. When you hover your cursor over the paragraph, a gray checkmark will appear. Click the mark to confirm the speaker. Select "Check all before this" as needed.