Skip to content

Fact · Dictation software

2.7times

Without linguistic errors, Dragon Medical One's mean word error rate was 2.7 times the Whisper-based model's rate in the pilot.

Checked against its source on October 16, 2026

What it measures
A ratio of two secondary results from a pilot of three physicians.
How it is calculated
1.34 / 0.50 = 2.68, rounded to 2.7
What it does not show
The ratio hides the overlap in the all-error result, where the difference was not significant. Read it with that result.

Calculated from

Cited on

Dragon's error rate was 2.7 times Whisper's without linguistic errors | Yap