VelociTap

Abstract

We present VelociTap: a state-of-the-art touchscreen keyboard decoder that supports a sentence-based text entry approach. VelociTap enables users to seamlessly choose from three word-delimiter actions: pushing a space key, swiping to the right, or simply omitting the space key and letting the decoder infer spaces automatically. We demonstrate that VelociTap has a significantly lower error rate than Google's keyboard while retaining the same entry rate. We show that intermediate visual feedback does not significantly affect entry or error rates and we find that using the space key results in the most accurate results. We also demonstrate that enabling flexible word-delimiter options does not incur an error rate penalty. Finally, we investigate how small we can make the keyboard when using VelociTap. We show that novice users can reach a mean entry rate of 41 wpm on a 40 mm wide smartwatch-sized keyboard at a 3% character error rate.

Investigating Effects of Post-Selection Feedback for Acquiring Ultra-Small Targets on Touchscreen

“…Although effective, these techniques sacrifice the simplicity and intuitiveness of direct touch, and largely reduce efficiency. For example, small-screen software keyboards based on such techniques [15,16] achieved only 25% of the input speed of direct input techniques [3,25].…”

Braille text entry on smartwatches

“…Finally, more recent studies try to solve the problem by providing statistical word models, without causing modifications on QWERTY layout. Examples of work with this strategy are WatchWriter [9], Velocitap [32] and the method analyzed in Turner et al [30]. These works seem to inspire current industry adopted methods, such as Google's WearOS default QWERTY keyboard, which relies heavily on tracing gestures for word composition and predictions.…”

HCI Research as Problem-Solving

“…Effectiveness is often easy to show for constructive papers, wherein the demonstration of a particular UI concept or technical realisation often is sufficient to show effectiveness (this is sometimes called an existence proof [16]). For instance, 'Velocitap' [48] shows that it is technically possible to perform sentence-level decoding of touchscreen text entry at interactive rates. Other papers tackling construction instead evaluate the technical performance of the solution or its user experience (there are various examples [29,31,35,50] [48], for instance, states that sentence-based decoding requires immense storage and processing power (a 2 GB language model; recognition on an eight-core server), but this is not a key concern.…”

“…This seems much more difficult for empirical problems, for which arguments about transfer are found in only a few studies. Confidence for the solution is addressed in multiple ways, including conducting several studies [7,24,48,50], using large samples [4], performing long-term follow-up [13], and reporting on implementation [13,44]. The sample covers this aspect quite well.…”