Fact · Voice and AI
0.45word error rate for Black speakers
In the study, Apple's system had a word error rate of 0.45 for Black speakers and 0.23 for white speakers.
Checked against its source on October 17, 2026
In the source's own words
“the WERs for black and white speakers are 0.45 (SE: 0.005) and 0.23 (SE: 0.003)”
Racial disparities in automated speech recognition, Allison Koenecke et al., Proceedings of the National Academy of Sciences, 2020, Results.
Open the original (DOI 10.1073/pnas.1915768117)- What it measures
- The authors describe Apple's system as the one with the worst overall performance among those they tested.
- What it does not show
- The text gives exact figures for two systems. The other systems' figures appear only in a chart in the paper.