Key points are not available for this paper at this time.
Speech technology promises to enable 'natural' experiences that remove the translation between a user's desires and a system's actions. While speech systems can translate speech into text, they cannot yet understand the vast range of users' intentions. The use of speech is also expanding from mobile contexts to environments like the living room. Given these shifts, key design questions emerge. How do we encourage natural speech while conveying scope? How do we provide feedback of understanding? What is the role of text-to-speech in this context? This paper presents findings from studying a prototype speech system enabling users to find movies and TV shows. We show the influence of a novel approach for conveying examples, a method for seamlessly integrating text and audio feedback, and share guidance on the use of text-to-speech in the living room.
Stifelman et al. (Sat,) studied this question.