Auditory Evoked Potentials (AEPs) have recently gained attention as a biometric feature that may improve security and address reliability shortfalls of other commonly-used biometric features. The objective of this thesis is to investigate the accuracy with which subjects can be automatically identified or authenticated with machine learning (ML) techniques using a type of AEP known as the speech-evoked frequency following response (FFR). Accordingly, the results show more accurate discrimination between FFRs from different subjects than what has been reported in past studies. The accuracy improvement is searched either by optimized hyperparameter tuning of the ML model or extracting new features from FFRs and feeding them as inputs to the model. Finally, the accuracy of authenticating subjects using FFRs is investigated using a "sheep vs. wolves" scenario. The results of this work shed more light on the potential of use of speech-evoked FFRs in biometric identification and authentication systems.

A central feature of conversation is that people take it in turns to speak. Typically speaker-listener roles are exchanged in a smooth and orderly fashion, with little or no gap or overlap. To date, within psychology only one comprehensive model of turn-taking has been proposed (Duncan, 1972). This model is cue based and suggests that discrete cues are responsible for the smooth management of conversation. There are, however, a number of fundamental shortcomings in the methodological and conceptual analysis that underpins this model. The aim of this thesis was to address these shortcomings for they have broader implications for our understanding of the turn exchange process. The methodology employed involved both the qualitative and quantitative micro-analysis of conversational data. To test the general significance of this analysis a more experimental approach, involving subjects judgements about particular sections of conversation, was employed. In order to put the generality question to the test, the investigations were based on different types of conversations - face-to-face conversations involving agreement and disagreement and telephone conversations involving travel enquiries and directory enquiries. The research carried out in this thesis has demonstrated that a wider range of information is exploited for turn-taking purposes than previously thought. The turn-taking cues Duncan identified could not provide an adequate explanation of how a smooth exchange of turns was actualised at a particular location. Two judgement studies demonstrated that whilst some conversations were managed by discrete cues as Duncan had suggested, others were not. Further investigations provided evidence that certain aspects of verbal content provide higher order and local information that is important for turn-taking. These investigations thus demonstrated that a cue based model of turn-taking is inadequate and emphasize the need for future work to provide precise explanations about how contextual factors are exploited in this process.

Research has shown that when a person who stutters self-discloses to a listener that he or she is a stutterer, this self-disclosure may positively impact the listeners' perceptions of the stuttering speaker. However, findings from previous studies have been limited in several ways. The purpose of this study was to further examine if listeners' perceptions of a stuttering speaker vary depending on whether or not the speaker self-discloses that they stutter, and if listener perceptions are subject to gender bias. We addressed limitations of prior studies by utilizing both male and female stuttering speakers who spoke directly to the viewer of the video, by balancing combinations of video viewings to account for potential effects of order, and by attempting to recruit a larger number of subjects. Participants (n =27) were randomly assigned to view two of the four possible videos (male self-disclosure, male no self-disclosure, female self-disclosure, and female no self-disclosure). After viewing both videos, participants immediately filled out a survey assessing their perceptions of the speakers' personality traits. Results for effects of self-disclosure achieved significance for "no difference" for traits of more intelligent, more unintelligent, and more unfriendly. These non-significant trends suggest self-disclosure might positively affect listener perceptions of a stuttering speaker, Results for effects of gender achieved significance for "no difference" for traits of more intelligent, more unintelligent, and more confident. In summary, results from this study show potential emerging trends that self-disclosure positively affects listeners' perceptions of stuttering speakers. / text

The literature suggests that self-disclosure of stuttering may positively impact the listener’s perception of persons who stutter. This phenomenon, although investigated with adults, has not been studied with regards to children who stutter. The purpose of this study was to examine the effects of self-disclosure on listener perceptions of children who stutter. Specifically, this study examined whether listener perceptions of a child speaker who stutters are contingent upon the presence or absence of a self-disclosure statement prior to the speaker initiating his or her monologue, and whether listener perceptions are susceptible to gender bias. Child participants (n = 71) between the ages of 6 years, 0 months and 12 years, 11 months were randomly assigned to view two of the four possible videos (male self-disclosure, male no self-disclosure, female self-disclosure, and female no self-disclosure). Directly following the viewing of both videos, participants completed a survey analyzing their perceptions of the speaker for various traits related to personality and intelligence. Results for effects of self-disclosure achieved significance for all ten questions. With regard to gender, there was a significant difference for all questions except ‘more unintelligent’ and ‘less distracted’; however, when the gender viewing possibilities were compared across the three distinct groups (distinguished by whether the video pairing included the male speaker only, the female speaker only, or both a male and female speaker), there was no significant difference found. This suggests that the difference across responses are only present when all three gender groups are collapsed, which further indicates that gender did not have a distinct impact on the responses to the questions. Additionally, there was no significant interaction between self-disclosure and gender, suggesting that these two factors have independent, un-related influence on listener perception. In summary, the present findings indicate that the use of self-disclosure may positively impact children’s perceptions of other children who stutter, and that these perceptions are not uniquely impacted by gender. / text

The purpose of this study was to examine how varying pause lengths in speech of people with aphasia (PWA) influences listeners' perceptions. The study specifically assesses listeners' perceptions of communicative effectiveness and speaker likability. Speech samples from six people with nonfluent or fluent aphasia were obtained from a previous study conducted by Harmon (2018). The speech samples were modified to create four sets of stimuli, including the original recordings, normalized within utterance pauses, normalized between utterance pauses, and normalized for both within and between utterance pauses. Forty listeners rated each of the speech samples based on the perceived communicative effectiveness and likability using a visual analog scale. Communicative effectiveness and likability ratings were significantly higher for the normalized within utterance and normalized within and between utterance conditions when compared to the baseline and normalized between utterance conditions. Both male and female listeners rated the recordings from nonfluent aphasic speakers lower than recording from speakers with fluent aphasia. Results of the study provide preliminary evidence that pauses in speech of PWA influence listeners' perceptions of communicative effectiveness and likability of the speaker. It is hoped that additional research regarding pause in speech of PWA will be conducted to determine if targeting pause in speech-language therapy will improve the communication of PWA.

The number of electric vehicles on the road is increasing rapidly every year. Due to the decreased sound produced by these vehicles at low speeds, there is significant concern that pedestrians and bicyclists will be at increased risk of vehicle collisions. This is particularly true for those with vision impairment who cannot rely on visual cues to alert them of an approaching vehicle. This thesis explores pedestrian aural detectability of six electric vehicle additive sounds produced by two additive sound systems: a modified version of the factory equipped system and a prototype exciter transducer-based system. All additive sounds and systems were first evaluated for regulatory compliance at stationary, 10 km/h, and 20 km/h conditions and then pedestrian detectability was assessed using 16 blind folded participants and on-road drive by tests. Participant drive by tests were then replicated using 3D sound field recordings played in a high-fidelity immersive reality lab. Results were used to verify the accuracy of lab environment and its potential applicability to future testing. The exciter transducer acoustic warning system was found to created more uniform sound levels on the passenger and drivers' sides of the vehicle than the factory system but produced lower sound levels on the front side of the vehicle. Additive sound modulation rate was not determined to be a key differentiator in pedestrian detectability and low frequency emphasis sounds were found to have the highest level of pedestrian detectability. As expected, vehicle speed played a critical role in participant detection with the 20 km/h speed condition producing higher average detection distances. The immersive reality lab was found to not replicate on-road environment however a perceived linear offset was observed between the two environments. / Master of Science / The number of electric vehicles on the road increases every year due to growing consumer demand for clean and sustainable transportation. Due to the decreased sound produced by these vehicles at low speeds there is significant concern that pedestrians and bicyclists will be at increased risk of vehicle collisions. This is particularly true for those with vision impairment who cannot rely on visual cues to alert them of an approaching vehicle. This thesis explores pedestrians' ability to detect six electric vehicle additive sounds produced by two sound systems: a modified version of the factory equipped system and a prototype system designed to produce uniform sound around the vehicle. All sounds and systems were evaluated see if they met current regulations at stationary, 10 km/h, and 20 km/h conditions. Pedestrians' ability to detect the vehicle was assessed using 16 blind folded participants and on-road tests where participants were asked to press a button when they heard an approaching vehicle. Participant drive by tests were then replicated using recordings taken on the same section of road and played in a lab environment. Results were used to see if the lab environment matched the results seen on the road. The prototype system created more uniform sound levels on the passenger and drivers' sides of the vehicle than the factory system but consistently produced lower sound levels on the front side of the vehicle. Sound modulation rate was not determined to be a key differentiator in pedestrian detectability and low frequency emphasis sounds were found to be the most easily detected by pedestrians. As expected, vehicle speed played a critical role in participant detection with the 20 km/h speed condition producing higher detection distances. The lab environment was found to not replicate on-road environment however similar offsets and sound ordering was observed between the two environments. Further work will be needed assess and correct this disagreement.

The demand for audiobooks is higher than ever--and increasing each year, surpassing the growth rate of electronic books, print books, and the US economy overall. Audiobooks have long been used to aid adolescent and struggling readers; however, recent studies have revealed that the majority of audiences consume audiobooks for the purpose of entertainment and relaxation. Despite this, much of the research available on audiobooks still applies to audiobooks as they are used in education. The purpose of this study was to understand the attitudes, opinions, and beliefs of people who love listening to audiobooks. The study was conducted using Q methodology, a behavioral research approach that combines quantitative and qualitative methods in order to measure audience perspectives. The results produced four factors: (a) Busy Bookworms, who listen because they do not have time to sit down and read; (b) Non-Readers, who enjoy audiobooks because they dislike reading physical books; (c) Experience Lovers, who listen so they can be immersed in a world of their own; and (d) Entertainment Seekers, who view audiobooks as a reward to motivate them to accomplish mundane daily tasks. The results of this study also revealed nuanced multitasking behaviors, as well as psychological gratifications for audiobook consumption.

The purpose of this study was to determine differences between the same subjects' music preferences at the elementary and high school levels, and the relationship between these findings and the following variables: peer preferences, musical training, excerpt familiarity, grade, gender, and race.

No description available.

Improving discourse is often targeted in aphasia treatment because it is an important skill for meaningful conversation and interaction. The aphasia literature offers a variety of methods to analyze and treat discourse impairments in aphasia, however, there is no true consensus on what the best method is for discourse analysis. Very few studies have utilized listener perception as a method to capture discourse-related changes in aphasia. However, many researchers in other areas of speech-language pathology (e.g. dysarthria, fluency) use listener perceptions and rating scales as a valid measure to assess connected speech. The overarching goal of this study is to determine whether people with aphasia (PWA) and naïve listeners perceive changes in discourse associated with conversational treatment. A questionnaire, the Discourse Rating Scale for Aphasia, was created based on three constructs of discourse analysis in aphasia: macrolinguistic, microlinguistic, and functional features. Six PWA and nine naïve listeners listened to 30-35 second speech samples obtained before and after conversational treatment and rated their judgments on the questionnaire. We examined the relationship between the ratings on the Discourse Rating Scale for Aphasia (DRSA) and standardized language tests to validate the items and rating scale. Additionally, we looked for descriptive pre and post differences within the data to determine whether the DRSA was sensitive to treatment. It was found that each item and total DRSA score correlated highly with standardized tests of language in aphasia. We did not find strong evidence for the DRSA’s sensitivity to treatment; however, we discuss the clinical implications of utilizing listener perception in the assessment of discourse in aphasia. / Communication Sciences