The Influence of the Auditory Environment on the Emotional Perception of Speech
Year: 2008
Type: article
Abstract: In this thesis the influence of the auditory environment on the emotional perception of speech in mediated communication is addressed. The motivation of this study is the development of techniques that enable suppression of environmental sound, with the goal to increasethe signal-to-noise ratio in communication devices, and thus improving speech intelligibility. Based on the commotion model of Schererand Zentner (2001), two routes were discerned in which the auditory environment is thought to affect the emotional percept of a sender. A direct route, changing the interpretation based on the auditory environment heard, and an indirect route, changing the way in which a sender speaks, and thereby the prosodic features that can be perceived and understood as symptoms of emotion. Two hypotheses were formulated: \n1. An auditory environment with a distinct emotional quality influences the perceived emotional quality of emotional speech\n2. Speech recorded in noisy environments is perceived less neutral when listened to outside the noisy context; i.e. more negative and more aroused. \nIn Experiment 1, stimuli were generated with two emotion induction methods, Velten-based and film-based. In Experiment 2, a selection of the stimuli was made. In experiments 3 and 4 hypotheses 1 and 2 were tested. Only for the first hypothesis evidence was found. To test hypothesis 1, emotional (positive, negative, neutral) utterances were played back in different (emotional) environments in Experiment 3. The environments significantly influenced the emotional perceptionof the sentences. In case of a positively perceived environment, utterances were in general perceived as more positive too, where as the negative environments lead to a more negative percept. Althoughon the emotional dimension of valence, the environmental sound seemed to displace the perceived valence of the utterance towards the valence of the environment, the two arousing noisy contexts, one rated more aroused than the other, lead to less aroused percepts when compared to the silent condition, suggesting a displacement away from the environment's emotional content. While confirmed in Experiment 4 for the neutral speech used in that experiment, the shift was not significant for the aroused Lombard speech, speech recorded in a noisy environment. The second hypothesis was rejected, therefore. Based on these results, it is not expected that suppressing the environmental sound in communication will lead to serious misunderstandings due to misinterpretation of a sender's emotion. Environmental noise does not seem to affect the perceived arousal nor the valence of the sender's emotion. Additionally, the attracting effect of the positive and negative contexts on the valence axis, the significant effect found for the environment resulting in a somewhat more neutral percept of someone's emotion, is not expected to lead to major misunderstandings either.
Author: Brouwers, MAJ Maarten
Institution: Philips (Netherlands)
Language: English
Topic: Emotion and Mood Recognition
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2105815390