Contribution
Command and Conquer: Towards a Robust Inventory of Voice Commands for HMI in Factory Halls
* Presenting author
Abstract:
In real-world contexts speech is infrequently produced without environmental noise, yet experimental paradigms frequently opt for optimal acoustic conditions (e.g., labs). Consequently, this limits the findings applicability to noisy environments like industrial workplaces. In industrial factories, the increase in automatization frequently results in more human-machine interaction often inefficiently executed manually by pressing buttons or entering commands via keyboard. As speech/voice interfaces have become more commonplace with the additional benefits of efficiency and convenience it does pose a challenge in workplaces with excessive noise (i.e., factories). Current speech science findings and models do not examine speech within the context of noise equating a large gap in present knowledge. The objective of the present study is to develop a speech/voice interface for noisy workplaces with a tailored inventory of commands resistant to environmental that are linguistically universally applicable and are easily learnable. Through a turn-based experimental design utilizing authentic, recorded industry noise environments, participants act as both speaker and listener with the accuracy and speed of responses and productions being recorded. By testing and evaluating different classes/combinations of sounds, we hope to establish a baseline the creation of this inventory of tailored commands. This study represents the beginning stages of this goal.