Keynote Speaker

Talk Title: In human-likeness we trust? Human-like speech synthesis and and the challenges of speech design and evaluation

Talk Abstract: Speech synthesis is a core part of any speech based conversational interface. It is arguably one of the most important aspects that we must consider as designers when creating speech based conversational machines. My keynote will explore how the frequent goal of human-likeness in synthesis influences user behaviour and perceptions of a speech agent’s capability as a dialogue partner (termed partner models), how the focus on emulating human voices may be misguided as well as how we can bridge the gaps between the speech technology and conversational user interfaces communities by jointly tackling key challenges in the evaluation of synthetic speech.

Brief Bio:  Benjamin Cowan is a Professor at  University College Dublin's School of Information and Communication Studies. His research focuses on understanding user interaction with speech-based conversational user interfaces. He is co-founder of the ACM Conversational User Interfaces conference (ACM CUI) and has published extensively on user interaction issues with speech interfaces within the HCI community. He is an Associate Editor of the International Journal of Human-Computer Studies, has served as CHI AC, and was one of the inaugural SCs of the Understanding People: Quantitative Methods subcommittee at CHI 2023.