The ability to hold listeners' attention and prevent habituation is a crucial design feature for both biological and human-made distress and alarm signals, but their intrusive nature also causes stress and annoyance. To investigate what makes vocalizations distracting and annoying, we developed Self-Steered Sound Synthesis-a new experimental paradigm for studying voice perception, in which untrained participants created fully synthetic yet realistic vocalizations by adjusting voice pitch, aspects of voice quality, and temporal structure of short vocalization sequences (dog barks, seagull calls, human shouts, etc.). Independent samples of listeners then evaluated the results. The primary determinants of vocal annoyance were as follows, from most to least important: (a) the amount of signal per unit of time, including the proportion of time spent vocalizing within a sequence and the proportion of unpleasant-sounding irregular phonation (nonlinear vocal phenomena); (b) the rate of acoustic events such as individual calls or sudden changes in voice quality; and (c) the unpredictable temporal structure of vocalization sequences, which was consistently increased in Self-Steered Sound Synthesis experiments but did not affect annoyance ratings. Thus, sequences of short vocal signals produced at a fast and/or unpredictable rate may be functionally optimal for alerting without undue annoyance, whereas continuous and harsh signals like baby cries cause maximum acoustic stress, making them impossible to ignore. With the Self-Steered Sound Synthesis method, complementary voice production and perception paradigms can be extended far beyond the range of vocal stimuli that participants can produce themselves, with wide-ranging applications in research on human and animal signals and sound design. (PsycInfo Database Record (c) 2026 APA, all rights reserved).
Anikin et al. (Mon,) studied this question.