The Impact of Stimulus Length in Tongue and Lip Movement Pattern Stability in Amyotrophic Lateral Sclerosis.

PURPOSE: This study aimed to investigate the effect of stimulus signal length on tongue and lip motion pattern stability in speakers diagnosed with amyotrophic lateral sclerosis (ALS) compared to healthy controls. METHOD: Electromagnetic articulography was used to derive articulatory motion patterns from individuals with mild (n = 27) and severe (n = 16) ALS and healthy controls (n = 25). The spatiotemporal index (STI) was used as a measure of articulatory stability. Two experiments were conducted to evaluate signal length effects on the STI: (a) the effect of the number of syllables on STI values and (b) increasing lengths of subcomponents of a single phrase. Two-way mixed analyses of variance were conducted to assess the effects of syllable length and group on the STI for the tongue tip (TT), tongue back (TB), and lower lip (LL). RESULTS: Experiment 1 showed a significant main effect of syllable length (TT, p < .001; TB, p < .001; and LL, p < .001) and group (TT, p = .037; TB, p = .007; and LL, p = .017). TB and LL stability was generally higher with speech stimuli that included a greater number of syllables. Articulatory variability was significantly higher in speakers diagnosed with ALS compared to healthy controls. Experiment 2 showed a significant main effect of length (TT, p < .001; TB, p = .015; and LL, p < .001), providing additional support that STI values tend to be greater when calculated on longer speech signals. CONCLUSIONS: Articulatory stability is influenced by the length of speech signals and manifests similarly in both healthy speakers and persons with ALS. TT stability may be significantly impacted by phonemic content due to greater movement flexibility. Compared to healthy controls, there was an increase in articulatory variability in those with ALS, which likely reflects deviations in speech motor control. SUPPLEMENTAL MATERIAL: https://doi.org/10.23641/asha.24463924.

Measuring Articulatory Patterns in Amyotrophic Lateral Sclerosis Using a Data-Driven Articulatory Consonant Distinctiveness Space Approach.

PURPOSE: The aim of this study was to leverage data-driven approaches, including a novel articulatory consonant distinctiveness space (ACDS) approach, to better understand speech motor control in amyotrophic lateral sclerosis (ALS). METHOD: Electromagnetic articulography was used to record tongue and lip movement data during the production of 10 consonants from healthy controls (n = 15) and individuals with ALS (n = 47). To assess phoneme distinctness, speech data were analyzed using two classification algorithms, Procrustes matching (PM) and support vector machine (SVM), and the area/volume of the ACDS. Pearson's correlation coefficient was used to examine the relationship between bulbar impairment and the ACDS. Analysis of variance was used to examine the effects of bulbar impairment on consonant distinctiveness and consonant classification accuracies in clinical subgroups. RESULTS: There was a significant relationship between the ACDS and intelligible speaking rate (area, p = .003; volume, p = .010), and the Amyotrophic Lateral Sclerosis Functional Rating Scale-Revised (ALSFRS-R) bulbar subscore (area, p = .009; volume, p = .027). Consonant classification performance followed a consistent pattern with bulbar severity, where consonants produced by speakers with more severe ALS were classified less accurately (SVM = 75.27%; PM = 74.54%) than the healthy, asymptomatic, and mild-moderate groups. In severe ALS, area of the ACDS was significantly condensed compared to both asymptomatic (p = .004) and mild-moderate (p = .013) groups. There was no statistically significant difference in area between the severe ALS group and healthy speakers (p = .292). CONCLUSIONS: Our comprehensive approach is sensitive to early oromotor changes in response due to disease progression. The preserved articulatory consonant space may capture the use of compensatory adaptations to counteract influences of neurodegeneration. SUPPLEMENTAL MATERIAL: https://doi.org/10.23641/asha.22044320.

Acoustic and Kinematic Methods of Indexing Spatiotemporal Stability in Children With Developmental Language Disorder.

PURPOSE: The spatiotemporal index (STI) is a standard metric for quantifying the stability and patterning of speech movements. The STI has often been applied to individual speech articulators, but an STI derived from the acoustic signal offers a composite and easily obtained measure that incorporates multiple components of the speech production complex. In this work, we examine the relationship between kinematic and acoustic STIs in children with and without developmental language disorder (DLD), with the aim of determining whether the acoustic and kinematic STIs reflect similar degrees of production variability. METHOD: A total of 85 children with DLD and with typical language development (or typically developing [TD] children), aged 4-8 years, were studied. In this methodological article, two experiments were conducted: one deliberately selected because group differences were observed in the kinematic STI (i.e., sentence production) and one in which there were no group differences in the kinematic STI (i.e., nonword production). These two experiments are representative of speech stability studies. The aim was to determine whether the acoustic STI (i.e., amplitude envelope) results aligned with those obtained via the kinematic STI (i.e., lip motion). RESULTS: In sentence production, most group differences aligned across kinematic and acoustic STI measures. The acoustic, but not the kinematic, STI showed higher variability in children with DLD compared with the 6-year-old TD group. In nonword production, neither the kinematic STI nor the acoustic STI differentiated children with DLD from TD children. In each experiment, the kinematic and acoustic STIs showed a moderate-to-strong correlation. CONCLUSIONS: The kinematic and acoustic STIs assess different components of speech movement patterning. However, the relationship between acoustic and kinematic spatiotemporal stability is strong in two tasks of varying linguistic complexity in children with and without DLD. These findings are promising for future experimental work in this area.

Tongue and Lip Acceleration as a Measure of Speech Decline in Amyotrophic Lateral Sclerosis.

PURPOSE: The goal of this study was to examine the efficacy of acceleration-based articulatory measures in characterizing the decline in speech motor control due to amyotrophic lateral sclerosis (ALS). METHOD: Electromagnetic articulography was used to record tongue and lip movements during the production of 20 phrases. Data were collected from 50 individuals diagnosed with ALS. Articulatory kinematic variability was measured using the spatiotemporal index of both instantaneous acceleration and speed signals. Linear regression models were used to analyze the relationship between variability measures and intelligible speaking rate (a clinical measure of disease progression). A machine learning algorithm (support vector regression, SVR) was used to assess whether acceleration or speed features (e.g., mean, median, maximum) showed better performance at predicting speech severity in patients with ALS. RESULTS: As intelligible speaking rate declined, the variability of acceleration of tongue and lip movement patterns significantly increased (p < 0.001). The variability of speed and vertical displacement did not significantly predict speech performance measures. Additionally, based on R2 and root mean square error (RMSE) values, the SVR model was able to predict speech severity more accurately from acceleration features (R2 = 0.601, RMSE = 38.453) and displacement features (R2 = 0.218, RMSE = 52.700) than from speed features (R2 = 0.554, RMSE = 40.772). CONCLUSION: Results from these models highlight differences in speech motor control in participants with ALS. The variability in acceleration of tongue and lip movements increases as speech performance declines, potentially reflecting physiological deviations due to the progression of ALS. Our findings suggest that acceleration is a more sensitive indicator of speech deterioration due to ALS than displacement and speed and may contribute to improved algorithm designs for monitoring disease progression from speech signals.

Acoustic Predictors of Ease of Understanding in Spanish Speakers With Dysarthria Associated With Parkinson's Disease.

PURPOSE: The purpose of this study was to examine selected baseline acoustic features of hypokinetic dysarthria in Spanish speakers with Parkinson's disease (PD) and identify potential acoustic predictors of ease of understanding in Spanish. METHOD: Seventeen Spanish-speaking individuals with mild-to-moderate hypokinetic dysarthria secondary to PD and eight healthy controls were recorded reading a translation of the Rainbow Passage. Acoustic measures of vowel space area, as indicated by the formant centralization ratio (FCR), envelope modulation spectra (EMS), and articulation rate were derived from the speech samples. Additionally, 15 healthy adults rated ease of understanding of the recordings on a visual analogue scale. A multiple linear regression model was implemented to investigate the predictive value of the selected acoustic parameters on ease of understanding. RESULTS: Listeners' ease of understanding was significantly lower for speakers with dysarthria than for healthy controls. The FCR, EMS from the first 10 s of the reading passage, and the difference in EMS between the end and the beginning sections of the passage differed significantly between the two groups of speakers. Findings indicated that 67.7% of the variability in ease of understanding was explained by the predictive model, suggesting a moderately strong relationship between the acoustic and perceptual domains. CONCLUSIONS: Measures of envelope modulation spectra were found to be highly significant model predictors of ease of understanding of Spanish-speaking individuals with hypokinetic dysarthria associated with PD. Articulation rate was also found to be important (albeit to a lesser degree) in the predictive model. The formant centralization ratio should be further examined with a larger sample size and more severe dysarthria to determine its efficacy in predicting ease of understanding.

Beyond Speech Intelligibility: Quantifying Behavioral and Perceived Listening Effort in Response to Dysarthric Speech.

PURPOSE: This study investigated whether listener processing of dysarthric speech requires the recruitment of more cognitive resources (i.e., higher levels of listening effort) than neurotypical speech. We also explored relationships between behavioral listening effort, perceived listening effort, and objective measures of word transcription accuracy. METHOD: A word recall paradigm was used to index behavioral listening effort. The primary task involved word transcription, whereas a memory task involved recalling words from previous sentences. Nineteen listeners completed the paradigm twice, once while transcribing dysarthric speech and once while transcribing neurotypical speech. Perceived listening effort was rated using a visual analog scale. RESULTS: Results revealed significant effects of dysarthria on the likelihood of correct word recall, indicating that the transcription of dysarthric speech required higher levels of behavioral listening effort relative to neurotypical speech. There was also a significant relationship between transcription accuracy and measures of behavioral listening effort, such that listeners who were more accurate in understanding dysarthric speech exhibited smaller changes in word recall when listening to dysarthria. The subjective measure of perceived listening effort did not have a statistically significant correlation with measures of behavioral listening effort or transcription accuracy. CONCLUSIONS: Results suggest that cognitive resources, particularly listeners' working memory capacity, are more taxed when deciphering dysarthric versus neurotypical speech. An increased demand on these resources may affect a listener's ability to remember aspects of their conversations with people with dysarthria, even when the speaker is fully intelligible.

Speaker Adaptation on Articulation and Acoustics for Articulation-to-Speech Synthesis.

Silent speech interfaces (SSIs) convert non-audio bio-signals, such as articulatory movement, to speech. This technology has the potential to recover the speech ability of individuals who have lost their voice but can still articulate (e.g., laryngectomees). Articulation-to-speech (ATS) synthesis is an algorithm design of SSI that has the advantages of easy-implementation and low-latency, and therefore is becoming more popular. Current ATS studies focus on speaker-dependent (SD) models to avoid large variations of articulatory patterns and acoustic features across speakers. However, these designs are limited by the small data size from individual speakers. Speaker adaptation designs that include multiple speakers' data have the potential to address the issue of limited data size from single speakers; however, few prior studies have investigated their performance in ATS. In this paper, we investigated speaker adaptation on both the input articulation and the output acoustic signals (with or without direct inclusion of data from test speakers) using the publicly available electromagnetic articulatory (EMA) dataset. We used Procrustes matching and voice conversion for articulation and voice adaptation, respectively. The performance of the ATS models was measured objectively by the mel-cepstral distortions (MCDs). The synthetic speech samples were generated and are provided in the supplementary material. The results demonstrated the improvement brought by both Procrustes matching and voice conversion on speaker-independent ATS. With the direct inclusion of target speaker data in the training process, the speaker-adaptive ATS achieved a comparable performance to speaker-dependent ATS. To our knowledge, this is the first study that has demonstrated that speaker-adaptive ATS can achieve a non-statistically different performance to speaker-dependent ATS.

Voice Onset Time in Early- and Late-Stage Amyotrophic Lateral Sclerosis.

PURPOSE: Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease that affects bulbar functions including speech and voice. Voice onset time (VOT) was examined in speakers with ALS in early and late stages to explore the coordination of the articulatory and phonatory systems during speech production. METHOD: VOT was measured in nonword /bap/ produced by speakers with early-stage ALS (n = 11), late-stage ALS (n = 6), and healthy controls (n = 13), and compared with speech performance decline (a marker of disease progression) in ALS. RESULTS: Overall comparison of the VOT values among the three groups showed a significant difference, F(2,27) = 11.71, p < .01. Speakers in late-stage ALS displayed longer voicing lead (negative VOT) than both healthy speakers and speakers in early-stage ALS. VOT was also significantly negatively correlated with speech performance (i.e., Intelligible Speaking Rate), r(15) = .74, p < .01. CONCLUSIONS: Speakers with more severe ALS showed greater occurrence of voicing lead and longer voicing lead. Findings show voicing precedes articulatory onset with disease progression in the production of bilabial stops, which suggests that the relative timing of coordination between the supralaryngeal structures and the phonatory system is affected in the late stage of ALS.

Influences of Methodological Decisions on Assessing the Spatiotemporal Stability of Speech Movement Sequences.

PURPOSE: The spatiotemporal index (STI) is a widely used approach for measuring speech pattern stability across multiple repetitions of a stimulus. In this study, we examine how methodological choices in the implementation of the STI (including the number of repetitions, length of stimuli, and parsing procedure) can affect its value. METHOD: To evaluate how each methodological decision affects the STI, we use a synthetic data framework that allows for the generation of random productions of the template phrase "Buy Bobby a Puppy" at different stability levels. Within this framework, we conduct three experiments: Experiment 1 investigates the effects of the number of repetitions, Experiment 2 investigates the effects of stimulus length, and Experiment 3 investigates the effects of parsing errors. RESULTS: In Experiment 1, we observed that STI values based on fewer repetitions will systematically underestimate larger repetition estimates. Experiment 2 showed that STI values will tend to be higher when calculated on longer (multimovement) stimuli independent of any differences in the stability of the underlying speech patterns. Finally, in Experiment 3, we showed that even minor parsing errors (≈ 10 ms) increase the value of the STI. CONCLUSIONS: The results of this study illustrate that even minor choices in the implementation of the STI can have a noticeable impact on the resulting value. These findings highlight the care that needs to be taken when designing studies and comparing STI values across studies to ensure that different STI values are capturing real differences in motion pattern stability rather than trivial methodological variation.

The Effects of Symptom Onset Location on Automatic Amyotrophic Lateral Sclerosis Detection Using the Correlation Structure of Articulatory Movements.

Purpose Kinematic measurements of speech have demonstrated some success in automatic detection of early symptoms of amyotrophic lateral sclerosis (ALS). In this study, we examined how the region of symptom onset (bulbar vs. spinal) affects the ability of data-driven models to detect ALS. Method We used a correlation structure of articulatory movements combined with a machine learning model (i.e., artificial neural network) to detect differences between people with ALS and healthy controls. The performance of this system was evaluated separately for participants with bulbar onset and spinal onset to examine how region of onset affects classification performance. We then performed a regression analysis to examine how different severity measures and region of onset affects model performance. Results The proposed model was significantly more accurate in classifying the bulbar-onset participants, achieving an area under the curve of 0.809 relative to the 0.674 achieved for spinal-onset participants. The regression analysis, however, found that differences in classifier performance across participants were better explained by their speech performance (intelligible speaking rate), and no significant differences were observed based on region of onset when intelligible speaking rate was accounted for. Conclusions Although we found a significant difference in the model's ability to detect ALS depending on the region of onset, this disparity can be primarily explained by observable differences in speech motor symptoms. Thus, when the severity of speech symptoms (e.g., intelligible speaking rate) was accounted for, symptom onset location did not affect the proposed computational model's ability to detect ALS.

MEG Sensor Selection for Neural Speech Decoding.

Direct decoding of speech from the brain is a faster alternative to current electroencephalography (EEG) speller-based brain-computer interfaces (BCI) in providing communication assistance to locked-in patients. Magnetoencephalography (MEG) has recently shown great potential as a non-invasive neuroimaging modality for neural speech decoding, owing in part to its spatial selectivity over other high-temporal resolution devices. Standard MEG systems have a large number of cryogenically cooled channels/sensors (200 - 300) encapsulated within a fixed liquid helium dewar, precluding their use as wearable BCI devices. Fortunately, recently developed optically pumped magnetometers (OPM) do not require cryogens, and have the potential to be wearable and movable making them more suitable for BCI applications. This design is also modular allowing for customized montages to include only the sensors necessary for a particular task. As the number of sensors bears a heavy influence on the cost, size, and weight of MEG systems, minimizing the number of sensors is critical for designing practical MEG-based BCIs in the future. In this study, we sought to identify an optimal set of MEG channels to decode imagined and spoken phrases from the MEG signals. Using a forward selection algorithm with a support vector machine classifier we found that nine optimally located MEG gradiometers provided higher decoding accuracy compared to using all channels. Additionally, the forward selection algorithm achieved similar performance to dimensionality reduction using a stacked-sparse-autoencoder. Analysis of spatial dynamics of speech decoding suggested that both left and right hemisphere sensors contribute to speech decoding. Sensors approximately located near Broca's area were found to be commonly contributing among the higher-ranked sensors across all subjects.

Predicting Montreal Cognitive Assessment Scores From Measures of Speech and Language.

Purpose This study examined the relationship between measurements derived from spontaneous speech and participants' scores on the Montreal Cognitive Assessment. Method Participants (N = 521) aged between 64 and 97 years completed the cognitive assessment and were prompted to describe an early childhood memory. A range of acoustic and linguistic measures was extracted from the resulting speech sample. A least absolute shrinkage and selection operator approach was used to model the relationship between acoustic, lexical, and demographic information and participants' scores on the cognitive assessment. Results Using the covariance test statistic, four important variables were identified, which, together, explained 16.52% of the variance in participants' cognitive scores. Conclusions The degree to which cognition can be accurately predicted through spontaneously produced speech samples is limited. Statistically significant relationships were found between specific measurements of lexical variation, participants' speaking rate, and their scores on the Montreal Cognitive Assessment.

Examining Listener Reaction Time in the Perceptual Assessment of Dysarthria.

OBJECTIVE: In the perceptual assessment of dysarthria, various approaches are used to examine the accuracy of listeners' speech transcriptions and their subjective impressions of speech disorder. However, less attention has been given to the effort and cognitive resources required to process speech samples. This study explores the relationship between transcription accuracy, comprehensibility, subjective impressions of speech, and objective measures of reaction time (RT) to further examine the challenges involved in processing dysarthric speech. PATIENTS AND METHODS: Sixteen listeners completed 3 experimental listening tasks: a sentence transcription task, a rating scale task, and an RT task that required responses to veracity statements. In each task, the speech stimuli included speech from 8 individuals with dysarthria. RESULTS: Measurements from the 3 tasks were significantly related, with a correlation coefficient of -0.94 between average RT and transcription-based intelligibility scores and -0.89 between RT and listener ratings of dysarthria. Interrater reliability of RT measurements was relatively low when considering a single person's response to stimuli. However, reliability reached an acceptable level when a mean was taken from 8 listeners. CONCLUSIONS: RT tasks could be developed as a reliable adjunct in the assessment of listener effort and speech processing.

Predicting Intelligibility Gains in Dysarthria Through Automated Speech Feature Analysis.

Purpose: Behavioral speech modifications have variable effects on the intelligibility of speakers with dysarthria. In the companion article, a significant relationship was found between measures of speakers' baseline speech and their intelligibility gains following cues to speak louder and reduce rate (Fletcher, McAuliffe, Lansford, Sinex, & Liss, 2017). This study reexamines these features and assesses whether automated acoustic assessments can also be used to predict intelligibility gains. Method: Fifty speakers (7 older individuals and 43 with dysarthria) read a passage in habitual, loud, and slow speaking modes. Automated measurements of long-term average spectra, envelope modulation spectra, and Mel-frequency cepstral coefficients were extracted from short segments of participants' baseline speech. Intelligibility gains were statistically modeled, and the predictive power of the baseline speech measures was assessed using cross-validation. Results: Statistical models could predict the intelligibility gains of speakers they had not been trained on. The automated acoustic features were better able to predict speakers' improvement in the loud condition than the manual measures reported in the companion article. Conclusions: These acoustic analyses present a promising tool for rapidly assessing treatment options. Automated measures of baseline speech patterns may enable more selective inclusion criteria and stronger group outcomes within treatment studies.

The relationship between perceptual disturbances in dysarthric speech and automatic speech recognition performance.

State-of-the-art automatic speech recognition (ASR) engines perform well on healthy speech; however recent studies show that their performance on dysarthric speech is highly variable. This is because of the acoustic variability associated with the different dysarthria subtypes. This paper aims to develop a better understanding of how perceptual disturbances in dysarthric speech relate to ASR performance. Accurate ratings of a representative set of 32 dysarthric speakers along different perceptual dimensions are obtained and the performance of a representative ASR algorithm on the same set of speakers is analyzed. This work explores the relationship between these ratings and ASR performance and reveals that ASR performance can be predicted from perceptual disturbances in dysarthric speech with articulatory precision contributing the most to the prediction followed by prosody.

Empirically Estimable Classification Bounds Based on a Nonparametric Divergence Measure.

Information divergence functions play a critical role in statistics and information theory. In this paper we show that a non-parametric f-divergence measure can be used to provide improved bounds on the minimum binary classification probability of error for the case when the training and test data are drawn from the same distribution and for the case where there exists some mismatch between training and test distributions. We confirm the theoretical results by designing feature selection algorithms using the criteria from these bounds and by evaluating the algorithms on a series of pathological speech classification tasks.