����ɫ��

What is a literary history of speech-to-text technology? This book project examines how American literary artists historically rendered folk speech into text while negotiating issues of race, gender, class, disability status, and national origin. Drawing on archival records, it rethinks fairness in automated speech recognition (ASR) systems. By framing speech-to-text remediation as editorial labour, the project challenges software engineering workflows that treat data labelling as low-paid, menial work—rather than as culturally informed and interpretive practice.

Bringing together design, AI, and technology, the project by Assistant Professor Setsuko Yokoyama calls for a more process-oriented approach to ASR development—one that recognises data labelling as care work. It highlights how listening practices are fluid and shaped by evolving cultural norms, which are often glossed over by mechanical speech processing. In doing so, it surfaces the overlooked human labour and cultural insight essential to creating genuinely antiracist speech technologies.