Perceptual susceptibility to acoustic manipulations in speaker discrimination

Abstract

Listeners' ability to discriminate unfamiliar voices is often susceptible to the effects of manipulations of acoustic characteristics of the utterances. This vulnerability was quantified within a task in which participants determined if two utterances were spoken by the same or different speakers. Results of this task were analyzed in relation to a set of historical and novel parameters in order to hypothesize the role of those parameters in the decision process. Listener performance was first measured in a baseline task with unmodified stimuli, and then compared to responses with resynthesized stimuli under three conditions: (1) normalized mean-pitch; (2) normalized duration; and (3) normalized linear predictive coefficients (LPCs). The results of these experiments suggest that perceptual speaker discrimination is robust to acoustic changes, though mean-pitch and LPC modifications are more detrimental to a listener's ability to successfully identify same or different speaker pairings. However, this susceptibility was also found to be partially dependent on the specific speaker and utterances.

Impaired generalization of speaker identity in the perception of familiar and unfamiliar voices.

“…The aim of these previous studies was to identify sets of salient acoustic features used by listeners to make inferences about a speaker; these studies can, however, also be interpreted as showing evidence for a lack of generalization across variability in vocal signals (in those cases, introduced by acoustic manipulations). For successful generalization, the effect of manipulations on one parameter should be compensated for with little impact on performance, as listeners are known to rely on a number of potentially speaker-specific acoustic cues when extracting identity-related information (Lavner et al, 2000; Sell, Suied, Elhilali, & Shamma, 2015). Although studies comparing speaker recognition across different languages have shown decreases in recognition accuracy when speech is presented in a language unfamiliar to the listener (see Introduction), there is nonetheless some retention of ability in the absence of intelligibility (i.e., listeners do not perform at floor).…”

Distributed acoustic cues for caller identity in macaque vocalization

“…Previous studies have suggested that both laryngeal and supralaryngeal components of vocalizations are critically involved in voice perception. Human listeners have difficulty in recognizing voices when either of these components is manipulated [32][33][34]. Macaques have demonstrated an ability to discriminate pitch [35][36][37], a component of the vocal source, and to use it for the caller discrimination [14].…”

Acoustic Correlates of Auditory Object and Event Perception: Speakers, Musical Timbres, and Environmental Sounds

“…Our current understanding of auditory object and event perception has been greatly advanced by the analysis of metrics that code differences among pairs of items. Dissimilarity data can come directly from behavioral ratings provided by participants (the more dissimilar two stimuli are, the higher the dissimilarity rating; Giordano et al, 2011), or from confusion rates in identification tasks (the more similar two stimuli are, the higher the degree of confusion; Giordano and McAdams, 2010; Sell et al, 2015). Such approaches provide rich data that often come in the form of a matrix representing how dissimilar all the possible pairs of items are within a set (or of interest for a particular study).…”

“…That is, not every cue that could potentially identify a given speaker needs to be used at the same time, nor are the same cues even used consistently by the same listener (Kreiman et al, 1992). Nonetheless, findings based on dissimilarity data and acoustic analyses suggest that speaker identification relates to fundamental frequency (determined by the vocal folds), formants in the frequency spectrum (determined by the vocal tract), and spectral slope (Matsumoto et al, 1973; Murry and Singh, 1980; Van Dommelen, 1990; Baumann and Belin, 2010; Sell et al, 2015). These are often simplified to glottal (source) and vocal tract (resonant) dimensions (Matsumoto et al, 1973; Baumann and Belin, 2010).…”