Thesis Format
Monograph
Degree
Master of Science
Program
Neuroscience
Supervisor
Johnsrude, Ingrid S.
Abstract
Our knowledge of the impact of background noise on speech intelligibility mostly pertains to linguistic content, which is only one aspect of speech. Prosodic features of speech conveying emotional information help one to understand an individual’s mental state better than words alone can. However, our ability to perceive such content in noise remains poorly understood. In this study, participants listened to semantically neutral noise-masked speech spoken with different emotional tones of voice and then completed word and emotion recognition tasks. As noise increased, the recognition of all emotions was better preserved than that of the words spoken, though most notably for negative-valence emotions (i.e. sadness, anger, and fear). I conclude that emotional information communicated through the prosody of the carrier voice is better preserved than the words spoken by that voice, when background noise is present. These findings suggest that perceiving emotional and linguistic content in speech may rely on separate cognitive mechanisms.
Summary for Lay Audience
In difficult listening conditions, like busy restaurants, if background sounds are loud enough, conversing with someone can be quite effortful—even if they’re right in front of you. Fortunately, speech contains additional pieces of information beyond the words spoken, which can be exploited to aid speech intelligibility. Notably, one’s tone of voice can be altered to express several emotions that complement the accompanying words. We tend to speak with an emotional tone matching the meaning of our words spoken (i.e. sad news with a sad tone), so identifying the emotional prosody could help you predict speech in noisy environments. For example, imagine hearing the sentence: “I __ my mother”, but a car honk completely masks the second word—a happy tone might bias you toward hearing a word like “saw”, but a sad tone might bias a different word, like “miss”.
For most of us, our brains do this automatically, but for hearing-impaired individuals, the necessary information is improperly encoded into the speech signal. Understanding speech becomes increasingly difficult for these individuals, and renders what they ultimately hear as entirely monotone and unemotional. These deficits reduce one’s ability to understand another individual’s mental state, which can be socially isolating and debilitating. Evidence suggests that perceiving spoken words and their accompanying emotional tones rely on separate cognitive processes, so research should address both aspects of speech to develop therapies best catered to help these individuals.
While the effect of different levels of background noise on speech intelligibility has been extensively studied, its effect on the recognition ability of vocalized emotions remains poorly understood. In my project, I investigated whether word and emotional tone recognition of speech are affected differently by background noise. Participants listened to semantically neutral noise-masked sentences, spoken with different emotional tones, and then completed word and emotion recognition tasks. As background noise got louder, all emotions were more accurately recognized than the words spoken, particularly for negative emotions (i.e. sadness, anger, and fear), indicating that the emotional tone content of speech is better preserved in the presence of background noise than linguistic content.
Recommended Citation
Nemiroff, Jonah, "The Recognition of Emotion from Speech in Noise" (2024). Electronic Thesis and Dissertation Repository. 10615.
https://ir.lib.uwo.ca/etd/10615
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.