Auditory Icons, Earcons, Spearcons, and Speech: A Systematic Review and Meta-Analysis of Brief Audio Alerts in Human-Machine Interfaces

Abstract

Auditory displays commonly are used in safety-critical domains and are a vital component of universal and inclusive design practices. Despite several decades of research on brief auditory alerts for representing status and processes in user interfaces, there is no clear heuristic guidance for which type(s) of auditory alerts should be preferred for designing interfaces. We used evidence synthesis (systematic review and meta-analysis) to examine the effectiveness of different types of brief audio alerts. We identified articles comparing auditory icons (real-world sounds with an ecological relationship to their referent), earcons (abstract sounds with no ecological relationship to their referent), spearcons (accelerated/compressed speech), and speech alerts. We used meta-analysis to compare alerts across five different outcomes: accuracy, reaction time, subjective ratings, workload, and dual-task interference. For accuracy and reaction time, results indicated speech, spearcons, and other types of alerts (usually hybrid, e.g., spearcons plus speech) were superior to auditory icons, which in turn were superior to earcons. Earcons also were inferior to all other options with respect to subjective ratings. Analyses generally suggested parity among alert types for workload and dual-task interference. Based on currently available evidence, it appears that speech, spearcons, and hybrid (e.g., spearcons plus speech) auditory alerts result in better performance than auditory icons and especially earcons. Still, high heterogeneity in our analyses cannot rule out a wide range of possible effects, and our analyses could not directly address some of the concerns that have been raised regarding speech-based alerts. These findings can help to guide the selection of brief audio alerts in interface design.

Speech, Nonspeech Audio, and Visual Interruptions of a Tracking Task: A Replication and Extension of Nees & Sampsell (2021)

“…Within categories of auditory alerts, we found evidence to suggest that speech alerts might be preferable to non-speech (tonal) alerts. This aligns with a recent meta-analysis [11] that found a general trend toward superiority of speech-based brief alerts across a variety of metrics. We also have made our research tools available for future researchers (see links in the Appendix section), as implementing a tracking task in online data-collection platforms is not easily accomplished with currently available tools.…”

“…This leaves open the possibility that speech sounds potentially could result in the preemptive effects on a visual task described in Auditory Preemption Theory. Yet other recent research [11] has suggested that speech sounds potentially are more effective than nonspeech sounds with respect to performance with alerts, which emphasizes the need to study any potential preemptive effects of speech alerts.…”