Liss, Julie

Purpose: This study aimed to evaluate how speech signal compression algorithms affect the acoustic and perceptual characteristics of dysarthric speech. As telepractice becomes more common in speech-language pathology, particularly in the assessment of dysarthria, understanding the impact of telecommunication compression on speech fidelity is essential for clinical decision making. Method: Speech samples were collected from 15 individuals with dysarthria, recorded locally and simultaneously via Zoom (Version 6.2.11). The Opus codec was used to process the locally recorded speech samples under fullband, wideband, and narrowband compression conditions. Acoustic measures were derived for all samples. Twenty experienced speech-language pathologists (SLPs) rated vocal quality and nasality for sustained vowels and rated articulatory precision and orthographically transcribed sentences across each condition. Results: Narrowband compression was associated with significantly degraded voice quality measures (e.g., harmonic-to-noise ratio, shimmer) and reduced transcription accuracy. Despite these changes in acoustics and intelligibility, ratings of articulatory precision, nasality, and vocal quality remained stable across conditions, suggesting SLPs could adjust for compression artifacts when making perceptual judgments. Conclusions: Speech compression, particularly in low-bandwidth (i.e., narrow) condition, impacts acoustic fidelity and intelligibility, though expert listeners maintain reliable perceptual judgments. These findings underscore the need to consider compression effects in telepractice and highlight the importance of developing optimized protocols for remote dysarthria assessment.

Purpose: Model interpretability is a critical requirement for deploying artificial intelligence (AI) applications in clinical speech settings, yet existing evaluation methods often overlook this aspect. The complex nature of speech features and the opaque decision-making processes of AI models underscore the need for a tailored framework to assess interpretability in speech-based clinical AI systems. This article proposes an initial framework composed of eight key factors to consider when evaluating the interpretability of clinical speech AI models. These factors are categorized into two groups: functional factors and clinician-centered factors. The functional factors, which can be assessed independently without user involvement, include faithfulness and computational efficiency. The clinician-centered factors include four established ones from existing literature (cognitive load, human–AI task performance, mental model, and user trust) and two new factors tailored to the unique demands of clinical speech applications: clinical understandability and decision relevance. We further suggest evaluation methods for each of the identified factors and propose modifying existing instruments such as the System Usability Scale and the Healthcare Systems Usability Scale to evaluate the two newly introduced factors. Conclusions: Our identified factors form an initial framework for evaluating the interpretability of speech-based clinical AI systems, supporting the effective integration of AI into clinical workflows. Future works include conducting an evaluation experiment using the proposed framework and refining it further based on the findings from the experiment.

Objectives/Goals: Aspiration causes or aggravates lung diseases. While bedside swallow evaluations are not sensitive/specific, gold standard tests for aspiration are invasive, uncomfortable, expose patients to radiation, and are resource intensive. We propose the development and validation of an AI model that analyzes voice to noninvasively predict aspiration. Methods/Study Population: Retrospectively recorded [i] phonations from 163 unique ENT patients were analyzed for acoustic features including jitter, shimmer, harmonic to noise ratio (HNR), etc. Patients were classified into three groups: aspirators (Penetration-Aspiration Scale, PAS 6–8), probable (PAS 3–5), and non-aspirators (PAS 1–2) based on video fluoroscopic swallow (VFSS) findings. Multivariate analysis evaluated patient demographics, history of head and neck surgery, radiation, neurological illness, obstructive sleep apnea, esophageal disease, body mass index, and vocal cord dysfunction. Supervised machine learning using five folds cross-validated neural additive network modelling (NAM) was performed on the phonations of aspirator versus non-aspirators. The model was then validated using an independent, external database. Results/Anticipated Results: Aspirators were found to have quantifiably worse quality of sound with higher jitter and shimmer but lower harmonics noise ratio. NAM modeling classified aspirators and non-aspirators as distinct groups (aspirator NAM risk score 0.528+0.2478 (mean + std) vs. non-aspirator (control) risk score of 0.252+0.241 (mean + std); p Discussion/Significance of Impact: We report the use of voice as a novel, noninvasive biomarker to detect aspiration risk using machine learning techniques. This tool has the potential to be used for the safe and early detection of aspiration in a variety of clinical settings including intensive care units, wards, outpatient clinics, and remote monitoring.

Abstract Background/Objective: Changes in speech can be detected objectively before and during migraine attacks. The goal of this study was to interrogate whether speech changes can be detected in subjects with post-traumatic headache (PTH) attributed to mild traumatic brain injury (mTBI) and whether there are within-subject changes in speech during headaches compared to the headache-free state. Methods: Using a series of speech elicitation tasks uploaded via a mobile application, PTH subjects and healthy controls (HC) provided speech samples once every three days, over a period of 12 weeks. The following speech parameters were assessed: vowel space area, vowel articulation precision, consonant articulation precision, average pitch, pitch variance, speaking rate and pause rate. Speech samples of subjects with PTH were compared to HC. To assess speech changes associated with PTH, speech samples of subjects during headache were compared to speech samples when subjects were headache-free. All analyses were conducted using a mixed-effect model design.Results: Longitudinal speech samples were collected from nineteen subjects with PTH (mean age=42.5, SD=13.7) who were an average of 14 days (SD=32.2) from their mTBI at the time of enrollment and thirty-one HC (mean age=38.7, SD=12.5). Regardless of headache presence or absence, PTH subjects had longer pause rates and reductions in vowel and consonant articulation precision relative to HC. On days when speech was collected during a headache, there were longer pause rates, slower sentence speaking rates and less precise consonant articulation compared to the speech production of HC. During headache, PTH subjects had slower speaking rates and altered vowel articulation compared to when they were headache-free. Conclusions: Compared to HC, subjects with acute PTH demonstrate altered speech as measured by objective features of speech production. For individuals with PTH, speech was altered during headache vs. when they were headache-free, suggesting that speech alterations were related to PTH and not solely due to the underlying mTBI.

Purpose Telemedicine, used to offset disparities in access to speech-language therapy, relies on technology that utilizes compression algorithms to transmit signals efficiently. These algorithms have been thoroughly evaluated on healthy speech; however, the effects of compression algorithms on the intelligibility of disordered speech have not been adequately explored. Method This case study assessed acoustic and perceptual effects of resampling and speech compression (i.e., transcoding) on the speech of 2 individuals with dysarthria. Forced-choice vowel identification and transcription tasks were utilized, completed by 20 naive undergraduate listeners. Results Results showed relative improvements and decrements in intelligibility, on various measures, based on the speakers' acoustic profiles. The transcoding of the speech compression algorithm resulted in an enlarged vowel space area and associated improvements in vowel identification for 1 speaker and a smaller vowel space area and decreased vowel identification for the other speaker. Interestingly, there was an overall decrease in intelligibility in the transcription task in this condition for both speakers. Conclusions There is a complex interplay between dysarthria and compression algorithms that warrants further exploration. The findings suggest that it is critical to be mindful of apparent changes in intelligibility secondary to compression algorithms necessary for practicing telemedicine. Supplemental Material https://doi.org/10.23641/asha.7291940

Although intelligibility deficits are common across various forms of dysarthria, it was hypothesized that the causes of intelligibility reductions depend on the ways the degraded signal challenges listener cognitive-perceptual strategies. This hypothesis was confirmed by analyzing the lexical boundary errors (LBE) elicited by 44 speakers with one of four forms of dysarthria (ataxic, hypokinetic, hyperkinetic, mixed spastic-flaccid). Error patterns from 60 listener transcripts revealed reliance on metrical stress to segment the 80 six-syllable phrases. However, reductions and disturbances in the acoustic cues to syllabic stress associated with each form of dysarthria resulted in different LBE patterns. Explanations for these unique patterns are found, in part, in the metrical structures of the dysarthric speech. For example, the temporal relationships among stressed and unstressed syllables varied widely in hyperkinetic speech, which ostensibly prevented listeners from optimally exploiting metrical stress as a cue for segmentation. The equal-even temporal structure of the spastic-flaccid speech resulted in the highest proportion of boundary insertions. The presence of interword pauses facilitated the assignment of word onset to strong syllables. Results will be presented relative to models of lexical segmentation and clinical implications for targets of remediation in the dysarthrias. [Work supported from NIH-NIDCD 5R01 DC6859.]

The detrimental effect of speaker variability on word recognition has been reported using a wide variety of tasks and stimuli. In this study, the use of dysarthric speech alongside unimpaired speech allowed us to examine the effect of indexical variability from a new perspective. We compared voice-specificity effects in words produced by healthy speakers and the same words produced by dysarthric patients suffering from Huntington’s disease (HD). The speech of persons with HD is particularly suited to this issue because of its idiosyncratic and variable manifestation. Reaction time and accuracy results provided confirmatory evidence for Luce et al.’s (2003) time-course hypothesis: The slower responses occasioned by processing dysarthric stimuli were accompanied with greater reliance on surface details of the stimuli. Thus, specificity effects can be seen as resulting from a coping strategy called upon when the system is presented with an input that is highly dissimilar from canonical lexical representations. In addition, we found some evidence that the time-course hypothesis holds even when intelligibility is controlled. This finding confirms that time, rather than processing effort, is the cause of specificity effects, with activation confined to abstract representations in an early stage and spreading to episodic traces later on.

Disturbances in speech rhythm are common in dysarthria, and deficit patterns vary among dysarthria subtypes. Attempts to quantify differences in disordered rhythm patterns have met with limited success. Here, we attempt to differentiate various forms of dysarthria using metrics designed to quantify linguistic rhythm on a stress-timed versus syllable-timed continuum. Speakers with hypokinetic, ataxic, hyperkinetic, or mixed flaccid-spastic dysarthria read sentences. Rhythm metrics were calculated for these utterances based on vocalic and consonantal interval durations. Differences among groups were discovered for %V (percentage of utterance duration comprised of vocalic intervals), VarcoV (rate-normalized standard deviation of vocalic interval durations), and nPVI-V (normalized pairwise variability index for vocalic intervals). Differences were independent of variable speaking rates among groups. All dysarthric groups produced rhythm patterns less stress-timed than the control American English speakers, with lower VarcoV and nPVI-V and higher %V. Those with flaccid-spastic dysarthria secondary to amyotrophic lateral sclerosis exhibited a syllable-timed pattern more extreme than even control Spanish speakers, indicating minimal differentiation of stressed and unstressed syllables. These results show that the various types of rhythmic disturbance accompanying dysarthria can be effectively described and distinguished within a stress-/syllable-timed continuum and relative to healthy speech.

High-density electroencephalography (EEG) was used to evaluate cortical activity patterns during the auditory processing of speech presented at various levels of degradation in a sentence verification task. 25 healthy participants listened to true-false sentences produced with one of 3 channel levels (1, 6, 16), ranging from unintelligible, moderately intelligible, and highly intelligible. Behavioral data were collected via button press (reaction time and accuracy) for each sentence. The analysis of cortical activation patterns includes 1) the identification of independent components that account for variations of activity associated with degradation levels and 2) the examination of associated event-related potentials. Statistical analyses will be performed on individuals' cortical activity, in tandem with behavioral data. Preliminary results reveal differences in the timing and magnitude of frontal and temporal lobe activation that coincide with task difficulty. Results of the present investigation will inform hypotheses about regions of interest for further investigation and will bear on models of connected speech perception, including the neurobiological underpinnings of individual differences and manifestations in a listener's understanding of degraded speech.