Nima Zargham
University of Bremen, Bremen, Germany
Advances in speech recognition, language processing and natural interaction have led to an increased industrial and academic interest. While the robustness and usability of such systems are steadily increasing, speech-based systems are still susceptible to recognition errors. This makes intelligent error handling of utmost importance for the success of those systems. In this work, we integrated anticipatory error handling for a voice-controlled video game where the game would perform a locally optimized action in respect to goal completion and obstacle avoidance, when a command is not recognized. We evaluated the user experience of our approach versus traditional, repetition-based error handling (N = 34). Our results indicate that implementing anticipatory error handling can improve the usability of a system, if it follows the intention of the user. Otherwise, it impairs the user experience, even when deciding for technically optimal decisions.
https://dl.acm.org/doi/abs/10.1145/3491102.3502115
The ACM CHI Conference on Human Factors in Computing Systems (https://chi2022.acm.org/)