Search for: All records

Word predictions in a text entry interface can help accelerate a user’s input. This may especially be true for users who have a slow input rate due to some form of motor-impairment. The choice of how many word predictions to offer in a text entry interface is an important design decision. In this work, we offered different number of word predictions in a keyboard where able-bodied users had to dwell on a key for one second to click it. We found participants’ text entry rate did not improve with increasing number of predictions. 

We investigate typing on a QWERTY keyboard rendered in virtual reality. Our system tracks users’ hands in the virtual environment via a Leap Motion mounted on the front of a head mounted display. This allows typing on an auto-correcting midair keyboard without the need for auxiliary input devices such as gloves or handheld controllers. It supports input via the index fingers of one or both hands. We compare two keyboard designs: a normal QWERTY layout and a split layout. We found users typed at around 16 words-per-minute using one or both index fingers on the normal layout, and about 15 words-per-minute using both index fingers on the split layout. Users had a corrected error rate below 2% in all cases. To explore midair typing with limited or no visual feedback, we had users type on an invisible keyboard. Users typed on this keyboard at 11 words-per-minute at an error rate of 3.3% despite the keyboard providing almost no visual feedback. 

Typing every character in a text message may require more time or effort than strictly necessary. Skipping spaces or other characters may be able to speed input and also reduce a user's physical input effort. This can be particularly important for people with motor impairments. In a large crowdsourced study, we found workers frequently abbreviated text by omitting mid-word vowels. We designed a recognizer optimized for noisy input where users often omit spaces and mid-word vowels. We show using neural language models for selecting training text and rescoring sentences improved accuracy. On noisy touchscreen data collected from hundreds of users, we found accurate abbreviated input was possible even if a third of characters were omitted. Finally, in a study where users had to dwell for a second on each key, sentence abbreviated input was competitive with a conventional keyboard with word predictions. After practice, users wrote abbreviated sentences at 9.6 words-per-minute versus word input at 9.9 words-per-minute. 

Making good letter or word predictions can help accelerate the communication of users of high-tech AAC devices. This is particularly important for real-time person-to-person conversations. We investigate whether per forming speech recognition on the speaking-side of a conversation can improve language model based predictions. We compare the accuracy of three plausible microphone deployment options and the accuracy of two commercial speech recognition engines (Google and IBM Watson). We found that despite recognition word error rates of 7-16%, our ensemble of N-gram and recurrent neural network language models made predictions nearly as good as when they used the reference transcripts.