Speech-to-text through respeaking

Translation has always been concerned with making content available and accessible for new audiences, but in recent years, accessibility as a practice and a principle has come into its own. The ability to access content and information is fundamental to a more equitable world, where every person is able to access the rights accorded to them as a human being. This article explores the practice of respeaking, where live subtitles are generated in real-time using speech recognition, as a place where digital translation and access meet. Drawing on professional practice, discussions with practitioners, and research conducted into respeaking, this work examines the versatility that respeaking has shown when faced with the changing landscape of translation, and the new and varied demands for access that respeakers encounter. Focusing on respeaking in non-broadcast settings (e.g. education, business meetings, cultural events, interviews), the article explores the process of respeaking, different audience groups who use this service, and varied ways that respeaking workflows can be embedded into intralingual, interlingual, multilingual and multimodal access provision. This versatility in practice reflects the resilience and reflexivity that respeakers themselves require in this changing world.