James Parker3
Automatic speech recognition is among the most well-known and widely discussed aspects of machine listening. Its technological history, mid-twentieth century onwards, reflects a shift from rule-based to statistical models of machine learning. Today, it is a part of everyday engagements with voice assistants. Critical perspectives on speech recognition include its history across geographies and languages, its deployment in military and border regimes, the intersections between visual impairment or non-literacy and the emergence of speech recognition, as well as endeavours to automate writing and textual production in the workplace that set grounds for the commodification of everyday speech. Due to its overlap with other forms of natural language processing, speech recognition can help understand the histories and politics of both linguistic data production as well as speech interfaces.