In the same 2024 study of university lecture recordings, Google's average English word error rate was about 6.9 times that of Whisper large-v2.

Checked against its source on October 17, 2026

What it measures: Both figures come from one study that ran eleven speech recognition services on the same English lecture recordings.
How it is calculated: 20.1 percent / 2.9 percent = 6.9
What it does not show: One study on lecture audio from one set of recordings. The ratio says nothing about a different language, a clean studio microphone or a later version of either service.

Calculated from