This study aimed to evaluate the efficacy of Apple AirPods pro (2nd generation) Live Listen feature in enhancing word recognition and memory retention among individuals with varying degrees of hearing loss, as determined by their Signal-to-Noise Ratio (SNR) loss. Utilizing a single-group experimental design, the research measured participants' performance on word recognition and memory retention tasks with and without the Live Listen feature. Statistical analysis, including paired t-tests and linear regression, revealed significant improvements in word recognition (from 81.8% to 94.4%) and memory retention (from 43.8% to 59.4%) scores when the Live Listen feature was activated. Moreover, a positive correlation between SNR loss and recognition score improvements suggested a greater benefit for those with higher levels of hearing loss. However, the relationship with memory retention improvements was less pronounced. These findings underscore the potential of the Live Listen feature as an effective assistive listening device, highlighting its importance in enhancing auditory experiences for individuals with hearing impairments and encouraging further research into personalized auditory assistance technologies in noisy healthcare environments.

Speech analysis for clinical applications has emerged as a burgeoning field, providing valuable insights into an individual's physical and physiological state. Researchers have explored speech features for clinical applications, such as diagnosing, predicting, and monitoring various pathologies. Before presenting the new deep learning frameworks, this thesis introduces a study on conventional acoustic feature changes in subjects with post-traumatic headache (PTH) attributed to mild traumatic brain injury (mTBI). This work demonstrates the effectiveness of using speech signals to assess the pathological status of individuals. At the same time, it highlights some of the limitations of conventional acoustic and linguistic features, such as low repeatability and generalizability. Two critical characteristics of speech features are (1) good robustness, as speech features need to generalize across different corpora, and (2) high repeatability, as speech features need to be invariant to all confounding factors except the pathological state of targets. This thesis presents two research thrusts in the context of speech signals in clinical applications that focus on improving the robustness and repeatability of speech features, respectively. The first thrust introduces a deep learning framework to generate acoustic feature embeddings sensitive to vocal quality and robust across different corpora. A contrastive loss combined with a classification loss is used to train the model jointly, and data-warping techniques are employed to improve the robustness of embeddings. Empirical results demonstrate that the proposed method achieves high in-corpus and cross-corpus classification accuracy and generates good embeddings sensitive to voice quality and robust across different corpora. The second thrust introduces using the intra-class correlation coefficient (ICC) to evaluate the repeatability of embeddings. A novel regularizer, the ICC regularizer, is proposed to regularize deep neural networks to produce embeddings with higher repeatability. This ICC regularizer is implemented and applied to three speech applications: a clinical application, speaker verification, and voice style conversion. The experimental results reveal that the ICC regularizer improves the repeatability of learned embeddings compared to the contrastive loss, leading to enhanced performance in downstream tasks.

Stress, depression, and anxiety are prevailing mental health issues that affect individuals worldwide. As the search for effective solutions continues, advancements in technology have led to the development of digital tools for stress identification and management purposes. The Cigna StressWaves Test (CSWT) is a publicly available stress analysis toolkit that claims to use “clinical-grade” artificial intelligence (AI) technology to evaluate individual stress levels through speech. To investigate their claim, this research stands as an independent validation study involving 60 participants over the age of 18. The primary objective of the study is to assess the repeatability and efficacy of the CSWT as a stress measurement tool. Key results indicate the CSWT lacks test-retest reliability and convergent validity. This implies that the CWST is not a repeatable tool and does not provide similar stress outcomes relative to an established measure of stress, the Perceived Stress Scale (PSS). These findings cast doubt on the accuracy and effectiveness of the CWST as a stress assessment tool. The public availability of the CSWT and the claim that it is a “clinical-grade” tool highlights concerns regarding premature deployment of digital health tools for stress management.

The brain uses the somatosensory system to interact with the environment and control movements. Additionally, many movement disorders are associated with deficits in the somatosensory sensory system. Thus, understanding the somatosensory system is essential for developing treatments for movement disorders. Previous studies have extensively examined the role of the somatosensory system in controlling the lower and upper extremities; however, little is known about the contributions of the orofacial somatosensory system. The overall goal of this study was to determine factors that influence the sensitivity of the orofacial somatosensory system. To measure the somatosensory system's sensitivity, transcutaneous electrical current stimulation was applied to the skin overlaying the trigeminal nerve on the lower portion of the face. After applying stimulation, participants' sensitivity was determined through the detection of the electrical stimuli (i.e., perceptual threshold). The data analysis focused on the impact of (1) stimulation parameters, (2) electrode placement, and (3) motor tasks on the perceptual threshold. The results showed that, as expected, stimulation parameters (such as stimulation frequency and duration) influenced perceptual thresholds. However, electrode placement (left vs. right side of the face) and motor tasks (lip contraction vs. rest) did not influence perceptual thresholds. Overall, these findings have important implications for designing and developing therapeutic neuromodulation techniques based on trigeminal nerve stimulation.

Stroke is the leading cause of long-term disability in the U.S., with up to 60% of strokescausing speech loss. Individuals with severe stroke, who require the most frequent, intense speech therapy, often cannot adhere to treatments due to high cost and low success rates. Therefore, the ability to make functionally significant changes in individuals with severe post- stroke aphasia remains a key challenge for the rehabilitation community. This dissertation aimed to evaluate the efficacy of Startle Adjuvant Rehabilitation Therapy (START), a tele-enabled, low- cost treatment, to improve quality of life and speech in individuals with severe-to-moderate stroke. START is the exposure to startling acoustic stimuli during practice of motor tasks in individuals with stroke. START increases the speed and intensity of practice in severely impaired post-stroke reaching, with START eliciting muscle activity 2-3 times higher than maximum voluntary contraction. Voluntary reaching distance, onset, and final accuracy increased after a session of START, suggesting a rehabilitative effect. However, START has not been evaluated during impaired speech. The objective of this study is to determine if impaired speech can be elicited by startling acoustic stimuli, and if three days of START training can enhance clinical measures of moderate to severe post-stroke aphasia and apraxia of speech. This dissertation evaluates START in 42 individuals with post-stroke speech impairment via telehealth in a Phase 0 clinical trial. Results suggest that impaired speech can be elicited by startling acoustic stimuli and that START benefits individuals with severe-to-moderate post-stroke impairments in both linguistic and motor speech domains. This fills an important gap in aphasia care, as many speech therapies remain ineffective and financially inaccessible for patients with severe deficits. START is effective, remotely delivered, and may likely serve as an affordable adjuvant to traditional therapy for those that have poor access to quality care.

Diffusion Tensor Imaging may be used to understand brain differences within PD. Within the last couple of decades there has been an explosion of learning and development in neuroimaging techniques. Today, it is possible to monitor and track where a brain is needing blood during a specific task without much delay such as when using functional Magnetic Resonance Imaging (fMRI). It is also possible to track and visualize where and at which orientation water molecules in the brain are moving like in Diffusion Tensor Imaging (DTI). Data on certain diseases such as Parkinson’s Disease (PD) has grown considerably, and it is now known that people with PD can be assessed with cognitive tests in combination with neuroimaging to diagnose whether people with PD have cognitive decline in addition to any motor ability decline. The Montreal Cognitive Assessment (MoCA), Modified Semantic Fluency Test (MSF) and Mini-Mental State Exam (MMSE) are the primary tools and are often combined with fMRI or DTI for diagnosing if people with PD also have a mild cognitive impairment (MCI). The current thesis explored a group of cohort of PD patients and classified based on their MoCA, MSF, and Lexical Fluency (LF) scores. The results indicate specific brain differences in whether PD patients were low or high scorers on LF and MoCA scores. The current study’s findings adds to the existing literature that DTI may be more sensitive in detecting differences based on clinical scores.

Children with cleft palate with or without cleft lip (CP+/-L) often demonstrate disordered speech. Clinicians and researchers have a goal for children with CP+/-L to demonstrate typical speech when entering kindergarten; however, this benchmark is not routinely met. There is a large body of previous research examining speech articulation skills in this clinical population; however, there are continued questions regarding the severity of articulation deficits in children with CP+/-L, especially for the age range of children entering school. This dissertation aimed to provide additional information on speech accuracy and speech error usage in children with CP+/-L between the ages of four and seven years. Additionally, it explored individual and treatment characteristics that may influence articulation skills. Finally, it examined the relationship between speech accuracy during a sentence repetition task versus during a single-word naming task.

Children with CP+/-L presented with speech accuracy that differed according to manner of production. Speech accuracy for fricative phonemes was influenced by severity of hypernasality, although age and status of secondary surgery did not influence speech accuracy for fricatives. For place of articulation, children with CP+/-L demonstrated strongest accuracy of production for bilabial and velar phonemes, while alveolar and palatal phonemes were produced with lower accuracy. Children with clefting that involved the lip and alveolus demonstrated reduced speech accuracy for alveolar phonemes compared to children with clefts involving the hard and soft palate only.

Participants used a variety of speech error types, with developmental/phonological errors, anterior oral cleft speech characteristics, and compensatory errors occurring most frequently across the sample. Several factors impacted the type of speech errors used, including cleft type, severity of hypernasality, and age.

The results from this dissertation project support previous research findings and provide additional information regarding the severity of speech articulation deficits according to manner and place of consonant production and according to different speech error categories. This study adds information on individual and treatment characteristics that influenced speech accuracy and speech error usage.

In many biological research studies, including speech analysis, clinical research, and prediction studies, the validity of the study is dependent on the effectiveness of the training data set to represent the target population. For example, in speech analysis, if one is performing emotion classification based on speech, the performance of the classifier is mainly dependent on the number and quality of the training data set. For small sample sizes and unbalanced data, classifiers developed in this context may be focusing on the differences in the training data set rather than emotion (e.g., focusing on gender, age, and dialect).

This thesis evaluates several sampling methods and a non-parametric approach to sample sizes required to minimize the effect of these nuisance variables on classification performance. This work specifically focused on speech analysis applications, and hence the work was done with speech features like Mel-Frequency Cepstral Coefficients (MFCC) and Filter Bank Cepstral Coefficients (FBCC). The non-parametric divergence (D_p divergence) measure was used to study the difference between different sampling schemes (Stratified and Multistage sampling) and the changes due to the sentence types in the sampling set for the process.

Transcranial Current Stimulation (TCS) is a long-established method of modulating neuronal activity in the brain. One type of this stimulation, transcranial alternating current stimulation (tACS), is able to entrain endogenous oscillations and result in behavioral change. In the present study, we used five stimulation conditions: tACS at three different frequencies (6Hz, 12Hz, and 22Hz), transcranial random noise stimulation (tRNS), and a no-stimulation sham condition. In all stimulation conditions, we recorded electroencephalographic data to investigate the link between different frequencies of tACS and their effects on brain oscillations. We recruited 12 healthy participants. Each participant completed 30 trials of the stimulation conditions. In a given trial, we recorded brain activity for 10 seconds, stimulated for 12 seconds, and recorded an additional 10 seconds of brain activity. The difference between the average oscillation power before and after a stimulation condition indicated change in oscillation amplitude due to the stimulation. Our results showed the stimulation conditions entrained brain activity of a sub-group of participants.