Designing conversational voice user interfaces

"Earl Grey, hot." We are closer than ever to the promised land of the Star Trek computer. There have been huge strides recently in the technology of speech recognition, but the new era of speech-enabled devices (such as those with Siri, Ok Google, or Amazon Alexa integration) is in its infancy. Cathy Pearl shares key design principles to help you design the best VUI possible.

Siri has been around since 2011, but the adoption of such technology has been slow. People became fed up with the frustrations of telephone-based speech recognition systems and have been reluctant to try again with mobile devices. But things are changing. The Amazon Echo was advertised at the Super Bowl. Apple has Cookie Monster showing off Siri’s timer feature. More people are getting comfortable with the idea of speaking to their phone. In a recent report by MindMeld, 45% of people who have used speech on their phones started in the last year alone. 62% of smartphone users are now using voice. Speech recognition technology has improved dramatically in the last five years, but that’s only half the solution. Crafting a VUI that works well and helps users without leaving them frustrated still requires careful design. Designing VUIs requires principles that differ from designing GUIs, including the fact that (unlike when someone taps a button on their phone) you’re not always 100% certain what the user really said. Cathy Pearl shares best practices for designing today’s VUIs, whether for mobile apps or home assistants like the Amazon Echo, and explores what “conversational” means and if it really matters when you’re just trying to help a user get something done. You’ll leave with practices for designing VUIs you can put into place immediately, including confirmation strategies, how to deal with speech recognition failures, and tips for prototyping, disambiguation, and keeping track of conversational context.