Modeling listener perception of speaker similarity in dysarthria

Do Acoustic Characteristics of Dysarthria in People With Parkinson's Disease Differ Across Languages?

PURPOSE

Cross-language studies suggest more similarities than differences in how dysarthria affects the speech of people with Parkinson's disease (PwPD) who speak different languages. In this study, we aimed to identify the relative contribution of acoustic variables to distinguish PwPD from controls who spoke varieties of two Romance languages, French and Portuguese.

METHOD

This bi-national, cross-sectional, and case-controlled study included 129 PwPD and 124 healthy controls who spoke French or Portuguese. All participants underwent the same clinical examinations, voice/speech recordings, and self-assessment questionnaires. PwPD were evaluated off and on optimal medication. Inferential analyses included Disease (controls vs. PwPD) and Language (French vs. Portuguese) as factors, and random decision forest algorithms identified relevant acoustic variables able to distinguish participants: (a) by language (French vs. Portuguese) and (b) by clinical status (PwPD on and off medication vs. controls).

RESULTS

French-speaking and Portuguese-speaking individuals were distinguished from each other with over 90% accuracy by five acoustic variables (the mean fundamental frequency and the shimmer of the sustained vowel /a/ production, the oral diadochokinesis performance index, the relative sound level pressure and the relative sound pressure level standard deviation of the text reading). A distinct set of parameters discriminated between controls and PwPD: for men, maximum phonation time and the oral diadochokinesis speech proportion were the most significant variables; for women, variables calculated from the oral diadochokinesis were the most discriminative.

CONCLUSIONS

Acoustic variables related to phonation and voice quality distinguished between speakers of the two languages. Variables related to pneumophonic coordination and articulation rate were the more effective in distinguishing PwPD from controls. Thus, our research findings support that respiration and diadochokinesis tasks appear to be the most appropriate to pinpoint signs of dysarthria, which are largely homogeneous and language-universal. In contrast, identifying language-specific variables with the speech tasks and acoustic variables studied was less conclusive.

Investigating Perceptual Subgroups in Speakers With Ataxic Dysarthria: An Auditory Free Classification Approach

PURPOSE

Ataxic dysarthria has presented with considerable heterogeneity in the presentation of speech characteristics. Converging evidence supports the existence of subgroups, specifically related to the instability and inflexibility of motor patterns as a possible explanation of this variability.

METHOD

To further examine the alignment of the speech characteristics of ataxic dysarthria with the instability/inflexibility framework, 23 graduate student listeners participated in an auditory free classification task and a guided classification task. Listeners grouped 15 speakers with ataxic dysarthria based on their judgment of the most salient perceptual characteristics during two speaking tasks: alternating motion rates (AMRs) and connected speech (one to two sentences). Listener ratings were then compared with a priori determinations of speakers who fit the instability subgroup profile and the inflexibility subgroup profile.

RESULTS

Results of both the free classification and guided classification listening paradigms provided supportive evidence of subgroups, particularly for the AMR task, in the context of strong inter- and intrarater reliability.

CONCLUSION

This study adds to the growing evidence of the existence of instability and inflexibility subgroups in ataxic dysarthria and serves as a proof of concept for use of the auditory free classification paradigm in dysarthria subgroup research.

Oromotor Nonverbal Performance and Speech Motor Control: Theory and Review of Empirical Evidence

This position paper offers a perspective on the long-standing debate concerning the role of oromotor, nonverbal gestures in understanding typical and disordered speech motor control secondary to neurological disease. Oromotor nonverbal tasks are employed routinely in clinical and research settings, but a coherent rationale for their use is needed. The use of oromotor nonverbal performance to diagnose disease or dysarthria type, versus specific aspects of speech production deficits that contribute to loss of speech intelligibility, is argued to be an important part of the debate. Framing these issues are two models of speech motor control, the Integrative Model (IM) and Task-Dependent Model (TDM), which yield contrasting predictions of the relationship between oromotor nonverbal performance and speech motor control. Theoretical and empirical literature on task specificity in limb, hand, and eye motor control is reviewed to demonstrate its relevance to speech motor control. The IM rejects task specificity in speech motor control, whereas the TDM is defined by it. The theoretical claim of the IM proponents that the TDM requires a special, dedicated neural mechanism for speech production is rejected. Based on theoretical and empirical information, the utility of oromotor nonverbal tasks as a window into speech motor control is questionable.

Perceptual Classification of Motor Speech Disorders: The Role of Severity, Speech Task, and Listener's Expertise

PURPOSE

The clinical diagnosis of motor speech disorders (MSDs) is mainly based on perceptual approaches. However, studies on perceptual classification of MSDs often indicate low classification accuracy. The aim of this study was to determine in a forced-choice dichotomous decision-making task (a) how accuracy of speech-language pathologists (SLPs) in perceptually classifying apraxia of speech (AoS) and dysarthria is impacted by speech task, severity of MSD, and listener's expertise and (b) which perceptual features they use to classify.

METHOD

Speech samples from 29 neurotypical speakers, 14 with hypokinetic dysarthria associated with Parkinson's disease (HD), 10 with poststroke AoS, and six with mixed dysarthria associated with amyotrophic lateral sclerosis (MD-FlSp [combining flaccid and spastic dysarthria]), were classified by 20 expert SLPs and 20 student SLPs. Speech samples were elicited in spontaneous speech, text reading, oral diadochokinetic (DDK) tasks, and a sample concatenating text reading and DDK. For each recorded speech sample, SLPs answered three dichotomic questions following a diagnostic approach, (a) neurotypical versus pathological speaker, (b) AoS versus dysarthria, and (c) MD-FlSp versus HD, and a multiple-choice question on the features their decision was based on.

RESULTS

Overall classification accuracy was 72% with good interrater reliability, varying with SLP expertise, speech task, and MSD severity. Correct classification of speech samples was higher for speakers with dysarthria than for AoS and higher for HD than for MD-FlSp. Samples elicited with continuous speech reached the best classification rates. An average number of three perceptual features were used for correct classifications, and their type and combination differed between the three MSDs.

CONCLUSIONS

The auditory-perceptual classification of MSDs in a diagnostic approach reaches substantial performance only in expert SLPs with continuous speech samples, albeit with lower accuracy for AoS. Specific training associated with objective classification tools seems necessary to improve recognition of neurotypical speech and distinction between AoS and dysarthria.

Residual Neural Network precisely quantifies dysarthria severity-level based on short-duration speech segments

Recently, we have witnessed Deep Learning methodologies gaining significant attention for severity-based classification of dysarthric speech. Detecting dysarthria, quantifying its severity, are of paramount importance in various real-life applications, such as the assessment of patients' progression in treatments, which includes an adequate planning of their therapy and the improvement of speech-based interactive systems in order to handle pathologically-affected voices automatically. Notably, current speech-powered tools often deal with short-duration speech segments and, consequently, are less efficient in dealing with impaired speech, even by using Convolutional Neural Networks (CNNs). Thus, detecting dysarthria severity-level based on short speech segments might help in improving the performance and applicability of those systems. To achieve this goal, we propose a novel Residual Network (ResNet)-based technique which receives short-duration speech segments as input. Statistically meaningful objective analysis of our experiments, reported over standard Universal Access corpus, exhibits average values of 21.35% and 22.48% improvement, compared to the baseline CNN, in terms of classification accuracy and F1-score, respectively. For additional comparisons, tests with Gaussian Mixture Models and Light CNNs were also performed. Overall, the values of 98.90% and 98.00% for classification accuracy and F1-score, respectively, were obtained with the proposed ResNet approach, confirming its efficacy and reassuring its practical applicability.

The Effects of Speech Compression Algorithms on the Intelligibility of Two Individuals With Dysarthric Speech

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.

Generalized Adaptation to Dysarthric Speech

PURPOSE

Generalization of perceptual learning has received limited attention in listener adaptation studies with dysarthric speech. This study investigated whether adaptation to a talker with dysarthria could be predicted by the nature of the listener's prior familiarization experience, specifically similarity of perceptual features, and level of intelligibility.

METHOD

Following an intelligibility pretest involving a talker with ataxic dysarthria, 160 listeners were familiarized with 1 of 7 talkers with dysarthria-who differed from the test talker in terms of perceptual similarity (same, similar, dissimilar) and level of intelligibility (low, mid, high)-or a talker with no neurological impairment (control). Listeners then completed an intelligibility posttest on the test talker.

RESULTS

All listeners benefited from familiarization with a talker with dysarthria; however, adaptation to the test talker was superior when the familiarization talker had similar perceptual features and reduced when the familiarization talker had low intelligibility.

CONCLUSION

Evidence for both generalization and specificity of learning highlights the differential value of listeners' prior experiences for adaptation to, and improved understanding of, a talker with dysarthria. These findings broaden our theoretical knowledge of adaptation to degraded speech, as well as the clinical application of training paradigms that exploit perceptual processes for therapeutic gain.

Listeners' attitudes toward accented talkers uniquely predicts accented speech perception

Listeners' sensitivity to indexical information influences their ability to perceive and remember speech, but it is less clear if listeners' subjective ratings of talker characteristics also impact speech perception ability. The present experiment tested the increase in variance accounted for by listeners' ratings of foreign accented talkers' manner of speaking and of the talkers themselves beyond the variance already accounted for by listeners' age, executive function, and hearing thresholds. Adding listeners' ratings significantly improved model fit, indicating that listeners' subjective experience of talker impacts speech perception accuracy along with objective listener characteristics such as hearing thresholds or executive function.