There Is a Racial Divide in Speech-Recognition Systems, Researchers Say - The New York Times

There Is a Racial Divide in Speech-Recognition Systems, Researchers Say - The New York Times

Racial discrepancies in voice-recognition AI

Speech-recognition software by leading technology companies is worse at identifying words spoken by African American people than those spoken by white Americans. Computer scientists tested algorithms by Amazon, Apple, Google, IBM and Microsoft with recordings of black people who live in predominantly black communities on the US East Coast and white people who live in mostly white areas in California. The systems made mistakes when transcribing speech 35% of the time for black speakers compared with 19% for white speakers.

New York Times | 5 min readReference: PNAS paper