AI transcription is powerful but not perfect—especially in noisy environments where speaker labels can get mixed up. This new feature lets you manually correct speaker labels on calls, ensuring coaching and insights apply to the right person every time.
Introducing Speaker Label Correction for Improved AI Transcriptions 👇
How to Correct Speaker Labels
Step 1: Open an Individual Call
From your Calls list, select the call where the speaker labels need correction.
Step 2: Flag the Call for Speaker Label Correction
At the top of the call view, click the three-dot menu (⋯) and select “Flag Call”.
Step 3: Select “Incorrect Speaker Labels”
In the flag options, choose “Incorrect speaker labels.”
Step 4: Assign Correct Speakers
Indicate who should be Speaker 1 and who should be Speaker 2 by selecting from the options.
If you’re unsure who is who, review the transcript to confirm who spoke first.
Step 5: Confirm and Reprocess
Click Confirm to submit the correction. Shilo will reprocess the call within about 30 seconds.
All summaries, coaching notes, and ratings will update to reflect the corrected speakers.
Additional Notes
Shilo’s AI gets speaker labels right about 98% of the time, so corrections will be rare but impactful.
This ensures you aren’t mistakenly graded or coached as the other speaker.