Evaluation of predictive text and speech inputs in a multimodal mobile route guidance application

Abstract

We present initial results from an evaluation of a multimodal interface for a mobile route guidance application providing public transport information in Finland. The application includes a range of input and output modalities, such as speech input and output, a fisheye GUI, and contextual predictive text input. The application has been in public pilot use for almost a year. In this paper we present results from a formal user evaluation, focusing on predictive text input and speech input and output. Speech input outperformed other input methods, even with high error rates and slow response times. However, the domain specific predictive text input was the preferred method, surpassing users' expectations.

User expectations and user experience with different modalities in a mobile phone controlled home entertainment system

“…Speech recognition was, in fact, quite robust, with overall accuracy of 93%, and 97% when OOV sentences are removed. However, as previous studies in similar multimodal conditions have shown, there is no clear correlation between the speech recognition accuracy and user experience [5], which is the case here as well. More importantly, our results show that grammar-based recognition, i.e., so-called restricted speech, can be used efficiently in this domain without user training.…”

“…A comparison to results from a previous study with the same experimental setup and a similar mobile application [5] shows that even if the expectations for speech input are mostly the same, the user experience of speech input observed in this study is superior, particularly when future use is considered. One reason for this is that in this scenario users found the speech input to provide additional benefit to the interaction that the other modalities could not easily offer, especially when issuing more complex commands.…”

“…For example, haptic feedback can provide silent, nonvisible feedback about the speech recognition process, without disturbing the users or filling up the display with graphical icons, saving space for the actual content. We have found that even simple haptic feedback can be very useful in multimodal mobile user interfaces [5]. On the other hand, research has shown that learning more complex haptic patterns requires training [3].…”

Eyes-free text entry on a touchscreen phone

“…Some studies explored speech as an input mechanism for mobile text entry [2,11]. General findings reveal that speech is an attractive alternative for text input.…”

Hype or Ready for Prime Time?

“…There have been some studies on multimodal interfaces for mobile devices or applications (e.g., Melto et al, 2008). Voice-enabled interfaces are widely available in modern cell phones, which have evolved from speaker-dependent name dialing that requires user enrollment of frequently dialed names, to speaker-independent features such as continuous digit dialing, command and control, and name dialing directly from the phone's contacts directory.…”