Cross-Cultural Perception of Spanish Synthetic Expressive Voices Among Asians

Ganapreeta Renunathan Naidu,Syaheerah Lebai Lutfi,Amal Azazi,Juan Martinez,Jaime Lorenzo-Trueba

doi:10.3390/app8030426

Ganapreeta Renunathan Naidu, Syaheerah Lebai Lutfi + Show 3 more

Open Access

https://doi.org/10.3390/app8030426

Copy DOI

Abstract

Nonverbal cues play a vital role in contributing to how emotions are perceived, especially by outgroups. In this study, a cross-cultural perception experiment of Spanish Synthetic Expressive Voices (SEV) was conducted to investigate the perception rate among different groups of Asians towards the SEV. Ten (10) subjects from each ethnic group namely Japanese, Chinese, Vietnamese, and Malaysians participated in this test. The subjects were required to listen to and categorize the SEV corpus which contains 260 utterances with 4 emotions (anger, happiness, sadness, and surprise) and the neutral speech in different intensities and durations. Overall, the results indicate that duration and intensity of speech plays a significant role in perception. This paper concludes that listeners’ perceptions are influenced by a speaker’s nonverbal expression and it is important that these features (duration and intensity of speech) are considered when modelling synthetic speech for artificial agents in real-time applications in a cross-cultural user environment.

Highlights

Nonverbal communication is defined as interaction without language
Studies on speech suggest that the perception of prosodic expressions has a universal component, further meta-analyses have attested that judges rated expressions from people of their own culture more accurately than they did with unfamiliar cultures [4,10]
The differences between variables—ethnic, duration, and intensity—were were analyzed with the univariate analysis of variance (ANOVA) while the multivariate analysis of analyzed with the univariate analysis of variance (ANOVA) while the multivariate analysis of variance variance (MANOVA) was applied to explore the interaction between the independent variables obtained from the experiment conducted for the “Simple4All” project [29]

Summary

Introduction

Vocal sounds that are not words, such as grunts or singing a wordless note, are nonverbal [1]. The human voice contains a rich source of nonverbal vocal expressive cues which come in different forms such as tone (prosodic expressions), breathing sounds, crying, hums, grunts, laughter, shrieks, and sighs (e.g., paralinguistic cues) [2,3]. Extensive reviews have established that the prosodic expressions of basic emotions—such as anger, fear, happiness, and sadness—are elicited by the acoustic patterns of cues related to pitch, intensity, voice quality, durations, and speed rates [4,5]. Studies on speech suggest that the perception of prosodic expressions has a universal component, further meta-analyses have attested that judges rated expressions from people of their own culture (in-group) more accurately than they did with unfamiliar cultures [4,10]. Interactions with in-group members (people from the same culture) are characterized by a sense of intimacy, familiarity, and trust while interactions with out-group members (people from a different culture) lack these qualities

Methods

Results

Conclusion