Recipes for building voice search UIs for automotive

Proceedings of the of the EACL 2014 Workshop on Dialogue in Motion (DM), pages 28–32, Gothenburg, Sweden, April 26-30 2014. c⃝2014 Association for Computational Linguistics Recipes for building voice search UIs for automotive Martin Labsky, Ladislav Kunc, Tomas Macek, Jan Kleindienst, Jan Vystrcil IBM Prague Research and Development Lab V Parku 2294/4, 148 00 Prague 4 Czech Republic {martin.labsky, ladislav kunc1, tomas macek, jankle, jan vystrcil}@cz.ibm.com Abstract In this paper we describe a set of tech- niques we found suitable for building multi-modal search applications for au- tomotive environments. As these ap- plications often search across different topical domains, such as maps, weather or Wikipedia, we discuss the problem of switching focus between different do- mains. Also, we propose techniques use- ful for minimizing the response time of the search system in mobile environment. We evaluate some of the proposed techniques by means of usability tests with 10 novice test subjects who drove a simulated lane change test on a driving simulator. We re- port results describing the induced driving distraction and user acceptance. 1 Introduction The task of designing mobile search user inter- faces (UIs) that combine multiple application do- mains (such as navigation, POI and web search) is significantly harder than just placing all sin- gle domain solutions adjacent to one another. We propose and evaluate a set of UI techniques use- ful for implementing such systems. The tech- niques are exemplified using a prototype multi- modal search assistant tailored for in-car use. The prototype supports several application domains in- cluding navigation and POI search, Wikipedia, weather forecasts and car owner’s manual. Fi- nally, we report usability evaluation results using this prototype. 2 Related Work Two examples of multi-modal search UIs for au- tomotive are the Toyota Entune1 and the Honda 1http://www.toyota.com/entune/ Link2. Both infotainment systems integrate a set of dedicated mobile applications including a browser, navigation, music services, stocks, weather or traffic information. Both use a tablet or a smartphone to run the mobile applications which brings the advantage of faster upgrades of the in- car infotainment suite. Home screens of these sys- tems consist of a matrix of square tiles that corre- spond to individual applications. The answers presented to the user should only contain highly relevant information, e.g. present- ing only points of interest that are near the cur- rent location. This is called conversational maxim of relevance (Paul, 1975). Many other lessons learned by evaluating in-car infotainment systems are discussed in (Green, 2013). In recent years, personal assistant systems like Siri (Aron, 2011), Google Now! (Google, 2013) and the Dragon Mobile Assistant (Nuance, 2013) started to penetrate the automotive environment. Most of these applications are being enhanced with driving modes to enable safer usage while driving. Dragon Mobile Assistant can detect whether the user is in a moving car and auto- matically switches to “Driver Mode” that relies on speech recognition and text-to-speech feed- back. Siri recently added spoken presentation of incoming text messages and voice mail, and it also allows to dictate responses. …

Content truncated to 500 words