In the 2024 study of university lecture recordings, IBM's average English word error rate was about 3.9 times that of Whisper large-v2.

Checked against its source on October 17, 2026

What it measures: Calculated from two numbers in the same study table. It covers English lecture recordings only, and it compares product versions from 2024.
How it is calculated: 11.2 / 2.9 = 3.86, rounded to 3.9
What it does not show: Derived from 11.2 and 2.9 percent, so the ratio inherits the limits of both numbers.

Calculated from