Case for a Voice-Internet: Voice Before Conversation

This position paper makes the case for constructing a voice-Internet; applications that run on voice activated personal agents (VAPAs) and that complement apps people run on their smartphones. I am promoting this position because screen readers will never really work. The content that screen readers read was designed to be seen, not heard. Text content relies on visual semantics for communication, it offers readers the ability to skim. Neither work with audio. This paper describes two studies that explore how VAPAs both do and do not meet the needs of current screen reader users. It then describes why now might be the perfect time to create a voice-Internet, noting that a technical platform already exists, that natural language processing technology is not yet ready for real conversation, and that enterprises show some willingness to create a new communication channel for their customers.