ABSTRACT
Short melodies are commonly linked to referents in jingles, ringtones, movie themes, and even auditory displays (i.e., sounds used in human–computer interactions). While melody associations can be quite effective, auditory alarms in medical devices are generally poorly learned and highly confused. Here, we draw on approaches and stimuli from both music cognition (melody recognition) and human factors (alarm design) to analyze the patterns of confusions in a paired-associate alarm-learning task involving both a standardized melodic alarm set (Experiment 1) and a set of novel melodies (Experiment 2). Although contour played a role in confusions (consistent with previous research), we observed several cases where melodies with similar contours were rarely confused – melodies holding musically distinctive features. This exploratory work suggests that salient features formed by an alarm’s melodic structure (such as repeated notes, distinct contours, and easily recognizable intervals) can increase the likelihood of correct alarm identification. We conclude that the use of musical principles and features may help future efforts to improve the design of auditory alarms.
Acknowledgements
We would like to thank Janet Kim, Fiona Manning, Glenn Paul, Olivia Podolak, Matthew Poon, and Jonathan Vaisberg for their assistance in data collection, and Jeanine Stefanucci for her assistance in exploring the ideas leading to this project.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
2. Demographic information could not be provided due to an unfortunate lab flooding in which we lost all hardcopies of participant information collected for this experiment before it could be saved electronically.
3. Flat alarms consisted of three tones 244 ms in length (including 25 ms rise/fall times) separated by 156 ms. The general structure of percussive alarms was the same with the exception of the individual tones having a rise time of 25 ms, followed by an immediate exponential decay for the remaining duration of the tone.
6. Previous explorations of IEC alarm learning classified individuals with least 1 year of musical training as “musically trained.”
7. We used SuperCollider (http://supercollider.github.io/) to shape pure tones (i.e., sine waves) into flat and percussive envelopes for 13 different pitches forming the one octave chromatic scale. These sequences used pitches iteratively selected (with replacement) from the original 13 tones. We then arranged these individual tones into sequences using Audacity (http://www.audacityteam.org) – a free sound-editing program. All tone sequences consisted of four 1-s sound clips, either all percussive or all flat, concatenated together to create a 4-s melody. Percussive tones were approximately 800 ms in length separated by approximately 150 ms. Flat tones were 745 ms in length separated by 200 ms. For additional technical details, see Schutz et al. (Citationin press).