Vocal caricatures reveal signatures of speaker identity

Abstract

What are the features that impersonators select to elicit a speaker's identity? We built a voice database of public figures (targets) and imitations produced by professional impersonators. They produced one imitation based on their memory of the target (caricature) and another one after listening to the target audio (replica). A set of naive participants then judged identity and similarity of pairs of voices. Identity was better evoked by the caricatures and replicas were perceived to be closer to the targets in terms of voice similarity. We used this data to map relevant acoustic dimensions for each task. Our results indicate that speaker identity is mainly associated with vocal tract features, while perception of voice similarity is related to vocal folds parameters. We therefore show the way in which acoustic caricatures emphasize identity features at the cost of loosing similarity, which allows drawing an analogy with caricatures in the visual space.

Deficits in Voice-Identity Processing: Acquired and Developmental Phonagnosia

“…Recognising voices at the individual level is a challenge for the perceptual and cognitive system. Each voice that we hear shares the same basic perceptual features across individuals (acoustic parameters such as pitch and timbre (Lavner et al, 2001;López et al, 2013)); and thus the brain is tasked with representing a unique voice in memory, by perceiving and representing often subtle differences in these features across individuals . Furthermore, it is not sufficient that we simply recognise a voice as familiar.…”

Familiarity and Voice Representation: From Acoustic-Based Representation to Voice Averages

“…For instance, it has been proposed that familiar speaker identity is mainly conveyed by three acoustic features: the fundamental frequency and the third and fourth formants (Lavner et al, 2001). More recently, a study using professional impersonators demonstrated that voice identity was mainly associated with acoustic features reflecting the anatomy of the vocal tract, such as the difference between the fourth and fifth formants (Lopez et al, 2013). Using trained-to-familiar voices, adaptation to identity was found only when the original configuration of the voice was preserved, suggesting that voice identity is represented in a complex pattern encompassing multiple acoustic information (Latinus and Belin, 2012).…”

Determination of Representativity for Bangla Vowel Perceptual Space

“…But, the invariant or representative acoustical features which activate the specific cortical circuit are still unknown [11]. Beside auditory model of vowel perception, vowel recognition for different languages are still improving by different non-auditory based techniques [12,13,14,15]. Among these techniques, determination of representative acoustical features of vowels and using these in perception purpose is a better technique as it can identify the key acoustical features of perception [15,16,17,18].…”

“…Among these techniques, determination of representative acoustical features of vowels and using these in perception purpose is a better technique as it can identify the key acoustical features of perception [15,16,17,18]. So, in both auditory and nonauditory models based speech perception, the representative acoustical features of vowel are important and a number of researches have already been conducted for searching the representative acoustical features for vowel perception [14,19,20,21,22,23].…”

“…in [21]. As formant dispersions or transitions affect on vowel perception process, these are also included in perceptual space in reports [22,23] [14]. Here, "D" refers dispersion of formant frequencies and the vector is frequently called "multidimensional acoustical vector".…”