Neurobrucellosis is an atypical and grave complication of brucellosis. It can present with a wide range of neurological symptoms, often mimicking other Central Nervous System (CNS) disorders, making diagnosis challenging. The present case report describes neurobrucellosis in a 47-year-old male who presented with atypical neurological symptoms, including depressed mood, dysarthric speech, increased sleepiness for 15 days, and altered sensorium for two days. Initial investigations, including routine Cerebrospinal Fluid (CSF) analysis, were non-contributory, and neuroimaging suggested acute encephalopathy with acute lacunar infarct. However, a high index of clinical suspicion due to his occupation as a farmer with cattle exposure led to advanced serological testing, which showed positive Brucella Immunoglobulin M (IgM), confirming Brucella infection. The patient was diagnosed with neurobrucellosis and responded favourably to a prolonged course of combination antibiotic therapy. This case underscores the need to include neurobrucellosis in the differential diagnosis of unexplained neurological symptoms, particularly in endemic areas or when there is a history of potential, even minimal, exposure.

Gestural feature extraction and multi-feature co-activation for dysarthric speech recognition

Raw acoustic-articulatory multimodal dysarthric speech recognition

PPFR-conformer for dysarthria speech recognition: from phoneme perception to feature refinement

Introduction: Deep Brain Stimulation (DBS) is an effective treatment for patients with advanced Parkinson’s disease (PD) who are refractory to optimized pharmacological therapy. Continuous stimulation is essential to maintain motor stability, and battery depletion can result in rapid clinical worsening. Recent studies confirm the long-term benefits of DBS on motor function, quality of life, and chronic neuromodulation.¹Clinical description: A 62-year-old male with a 12-year history of PD (akinetic-rigid subtype) initially presented with freezing, bradykinesia, and early morning rigidity. Despite treatment with pramipexole, entacapone, multiple levodopa formulations, clonazepam, tramadol, melatonin, and amantadine, motor symptoms remained refractory. In 2022, he underwent bilateral DBS implantation in the subthalamic nucleus (STN), resulting in significant rigidity improvement. After generator battery depletion, the patient experienced progressive return of symptoms: worsening bradykinesia, chronic pain, and loss of independence in daily activities. Neurological examination showed severe bradykinesia, marked hypomimia, dysarthric speech, mild asymmetric tremor, and freezing. Cranial CT confirmed correct lead placement. The patient underwent generator replacement, programming was reestablished using standard bilateral STN parameters: 130 Hz, 60 µs pulse width, 2.8 V (left) and 3.0 V (right), in monopolar configuration (contacts 1–2 left, 9–10 right). Immediate postoperative improvement was noted, with reduced rigidity and bradykinesia. At one-month follow-up, the patient reported complete pain resolution, functional recovery, improved sleep, and high satisfaction.Discussion: This case illustrates the essential role of continuous DBS in motor control for advanced PD. Studies report sustained improvements in UPDRS-III and PDQ-8 scores after DBS¹. Patient selection based on clinical and imaging profiles remains key to optimal outcomes.² Generator replacement is safe and effective, with low complication rates.³Conclusions: DBS generator replacement led to rapid functional recovery and quality of life improvement. This case reinforces the need for structured follow-up, early battery monitoring, and individualized DBS care.2,3

A Novel Hybrid Deep Learning Model Optimized by Pelican Algorithm for Robust Dysarthric Speech Classification with Multi-feature Fusion and Attention-GAN

Dysarthria, characterised by slurred speech, is a hallmark of many neurological disorders and brain trauma. Clinical assessment requires an audio-visual investigation by a trained healthcare expert, who evaluates criteria such as respiration, phonation, articulation, resonance, and prosody during speech. Quantitative assessment of dysarthria is challenging due to its complexity, variability, and the subjective nature of human-observation-based scoring methods. We present a novel machine-learning framework using transformers for stratifying and monitoring patient speech. Our framework integrates a wav2vec 2.0 model, pre-trained on raw speech data from healthy individuals. To reduce reliance on speaker-specific characteristics and effectively manage the intrinsic intra-class variability of dysarthric speech, we employ a contrastive learning strategy with a multi-task objective: cross-entropy loss for classifying dysarthria severity, and triplet margin loss to ensure latent embeddings are grouped by severity rather than by speaker. This Speaker-Agnostic Latent Regularisation (SALR) framework provides an objective, accessible, and cost-effective alternative to traditional assessments. On the UA-Speech dataset, SALR achieved 70.5% accuracy and 59.2% F1 using leave-one-subject-out cross-validation—a 16.5% absolute (30% relative) improvement over prior benchmarks. Explainability analysis indicates that our multi-task objective enhances the ordinal structure of the latent space, reducing dependence on speaker-specific cues and demonstrating robustness and generalisability. In conclusion, this proof-of-concept study demonstrates the potential of the SALR framework for speaker-independent dysarthria severity classification, with potential implications for broader clinical applications in automated dysarthria assessments.

Automatic Speech Recognition (ASR) for a typical speech, such as dysarthria, presents a significant challenge due to high acoustic variability, which often leads to failures in standard models. This challenge is further compounded when implementation is targeted for edge devices with limited computational resources, memory, and power. The need for model architectures that are not only accurate but also highly efficient (lightweight) is crucial for realizing on-device ASR systems with low latency. This research focuses on exploring modern deep learning architectures to address these two primary challenges: accuracy in dysarthric speech and computational efficiency. The study aims to implement and evaluate three efficient models—MobileNetV3Small, EfficientNetB0, and NASNetMobile—on the UASpeech and TORGO datasets. The methodology involves extracting Mel-Frequency Cepstral Coefficients (MFCC) features, which are visualized as spectrograms and subsequently classified using a transfer learning approach. Experimental results show that the MobileNetV3Small model achieved the highest performance on the UASPEECH dataset, attaining a uniform score of 97,8 % for accuracy. This study concludes that lightweight CNN architectures like MobileNetV3Small are highly effective for dysarthric speech classification and demonstrate the feasibility of developing robust and practical ASR systems for resource-constrained environments.

This article presents the main correctional-developmental, pedagogical opportunities and technologies for developing initial reading skills in preschool children with dysarthria speech impairment.

Dysarthric speech recognition faces significant challenges of acoustic variability and data scarcity, and this study proposes a robust system by integrating generative adversarial network enhancement and large language model correction to address these issues effectively. The system employs three key components, including a multimodal recognition core that combines whisper‐medium encoder with LoRA‐fine‐tuned Llama‐3.1‐8B for end‐to‐end acoustic‐to‐semantic mapping, an improved CycleGAN module that generates synthetic dysarthric speech through Inception‐ResNet fusion blocks, and an intelligent error correction mechanism using N‐best hypothesis reranking with semantic constraints. Experiments on the UA‐Speech dataset show that the complete system achieves a 20.61% word error rate representing a 73.9% relative improvement over traditional end‐to‐end transformer automatic speech recognition. Under very low intelligibility conditions it maintains a 48.69% word error rate demonstrating robust recognition for severe pathological speech. Ablation studies validate each module's effectiveness, providing significant advances for dysarthric patient communication technologies.

Here, we investigated how intelligibility is impacted in underappreciated, highly complex, but real-world communication scenarios involving two clinical populations-when the speaker has dysarthria and the listener has hearing loss, in noisy everyday environments. As a second aim, we examined the potential for modern noise reduction to mitigate the noise burden when listeners with hearing loss are attempting to understand a speaker with dysarthria. Thirteen adults with sensorineural hearing loss (SNHL) listened and transcribed dysarthric speech under three processing conditions: quiet, noise, and noise reduced. The intelligibility scores of listeners with SNHL were compared with previously reported data collected from adults without hearing loss (Borrie et al., 2023). Listeners with SNHL performed significantly poorer than typical-hearing listeners when listening to speech produced by a speaker with dysarthria-an intelligibility disadvantage that was exacerbated when background noise was present. However, it was also found that a time-frequency-based noise reduction technique was able to effectively restore the intelligibility of dysarthric speech in noise to approximate levels in quiet for listeners with hearing loss. The results highlight the substantial intelligibility burden placed upon a communication dyad consisting of a speaker with dysarthria and a listener with hearing loss, when background noise is present. Given the etiologies of dysarthria and hearing loss, and presence of noise in many everyday communication environments, this scenario is not uncommon. As such, these results are an important first step toward understanding the challenges experienced when communication disorders interact. The finding that noise reduction techniques can mitigate much of the noise burden provides a promising future direction for research that seeks to manage communication with two clinical populations.

Despite the prevalence of bilingualism, research on the understanding of disordered speech has focused almost exclusively on monolingual populations. Hypothesis-driven studies with dysarthric speech have revealed that greater vocabulary knowledge and working memory support understanding in monolingual listeners. However, whether these explanatory models generalize to bilinguals, who differ in both cognitive and linguistic profiles, is unknown. This study examined whether bilingualism affords a perceptual advantage in understanding dysarthric speech, and whether working memory and vocabulary knowledge contribute to that advantage. Ninety-four listeners, categorized as monolinguals, early bilinguals, or late bilinguals, completed a speech perception task where they transcribed phrases produced by speakers with dysarthria. They also completed working memory and vocabulary assessments. Relative to monolingual and late bilinguals, early bilinguals had equivalent working memory scores, lower vocabulary scores, and reduced intelligibility scores when perceiving dysarthric speech. Vocabulary knowledge, but not working memory, predicted intelligibility scores across all groups. A post hoc correlation analysis within the early bilingual group further revealed that an earlier age of exposure to English was associated with higher intelligibility scores, suggesting that age of language exposure plays a critical role in shaping the linguistic systems that support perception of disordered speech. These findings underscore the importance of vocabulary knowledge and language experience, over working memory, in facilitating understanding of disordered speech. They also highlight the need to refine models of disordered speech perception to account for variability across listener populations, in order to more fully capture the relative contributions of cognitive and linguistic mechanisms. https://doi.org/10.23641/asha.30220360.

Severity Classification of Dysarthric Speech using Soft Sets

Enhanced early detection of dysarthric speech disabilities using stacking ensemble deep learning model

The two predominating subtypes of late-onset GM2 gangliosidosis are late-onset Tay-Sachs (LOTS) and late-onset Sandhoff disease (LOSD). Due to shared deficiencies of ß-hexosamindase A and significant clinical overlap, the two diseases have been considered indistinguishable. However, a growing body of evidence supports the notion of several distinctions between the two diseases. In this study, we highlight these distinctions through the cross-sectional evaluation of 27 late-onset GM2 gangliosidosis participants. Twenty-one participants with LOTS and 6 with LOSD were included in this study. We performed physical examinations alongside assessments for gait, balance, muscle strength, ataxia, nerve conduction velocities, and analyzed brain magnetic resonance imaging. Lower limb weakness (95% in LOTS, 100% in LOSD) and later development of upper limb weakness (90% in LOTS, 83% in LOSD) was highly prevalent in both cohorts. Accompanying gait disturbances, balance issues, and dysmetria (as assessed by the brief ataxia rating scale [BARS]) were also prevalent in both cohorts. Strength testing for the quadriceps and hamstrings demonstrated weakness in both cohorts, primarily impacting extensor muscles. Supratentorial gray and white matter volumes in both cohorts were similar to normative data. In contrast, BARS scores for dysarthria and oculomotor dysfunction were present and heterogenous in LOTS participants and absent in LOSD participants. 24% of LOTS participants and none of the LOSD participants had a history of neuropsychiatric symptoms. Cerebellar volume including lobules V and VI were lower in LOTS compared to LOSD and normative data. However, length dependent sensory neuropathy was present in all LOSD participants but absent in LOTS participants. Dysfunction of the posterior cerebellum (lobules VI, VII, and IX) has been shown to cause cerebellar cognitive affective syndrome (CCAS), that includes cognitive and behavioral disturbances. Furthermore, cerebellar dysfunction of lobules V and VI has been linked to dysarthric speech, and dysfunction of the posterior cerebellum has been linked to oculomotor symptoms. The finding of low cerebellar lobule volumes in LOTS, suggests the distinctive features of the LOTS phenotype are related to cerebellar dysfunction. However, the sensory symptoms unique to LOSD remains a mystery. The molecular and biochemical basis for the dichotomy between the LOTS and LOSD phenotypes requires further investigation.

The assessment of dysarthria severity directly reflects the progression of a patient’s condition and serves as a crucial baseline for developing targeted intervention programs. Emotional characteristics embedded in dysarthric speech not only record the emotional state of patients but also assist clinicians in understanding their mental health status and advancing subsequent treatment. We innovatively designed a multi-scale feature fusion module that integrates speech emotion features with intelligibility characteristics using Fisher vector encoding, thereby enhancing the richness of input features in automated dysarthria assessment systems. In this study, we conducted multiple comparative experiments using different acoustic features and deep learning techniques. The results demonstrate that our multi-scale feature approach achieves an accuracy of 98.56% with the deep neural networks (DNNs) classification model and an impressive 96.13% with the support vector machine (SVM). These findings validate the effectiveness of the multi-scale feature fusion approach in dysarthria severity level assessment and provide new perspectives for the medical diagnosis of dysarthria.

Dysarthria frequently occurs in individuals with disorders such as stroke, Parkinson’s disease, cerebral palsy, and other neurological disorders. Well-timed detection and management of dysarthria in these patients is imperative for efficiently handling the development of their condition. Several previous studies have concentrated on detecting dysarthria speech using machine learning-based methods. However, the false positive rate is high due to the varying nature of speech and environmental factors such as background noise. Therefore, in this work, we employ a model based on the Swin transformer (ST), namely DSR-Swinoid. Firstly, the speech is converted into mel-spectrograms to reflect the maximum patterns of voice signals. Despite the ST’s initial aim to effectively extract the local and global visual features, it still prioritizes global features. Meanwhile, in mel-spectrograms, the specific gaps due to slurred speech are considered. Therefore, our objective is to improve the ST’s capacity for learning local features by introducing 4 modules: network for local feature capturing (NLF), convolutional patch concatenation, multi-path (MP), and multi-view block (MVB). The NLF module enriches the existing Swin transformer by enhancing its capability to capture local features effectively. MP integrates features from different Swin phases to emphasize local information. In the meantime, the MVB-ST block surpasses classical Swin blocks by integrating diverse receptive fields, focusing on a more comprehensive extraction of local features. Investigational outcomes explain that the DSR-Swinoid attains the best exactness of 98.66%, exceeding the outcomes by existing methods.

The GRAM-KAN approach: dysarthria speech identification with easy call corpus

Analysis and development of clinically recorded dysarthric speech corpus for patients affected with various stroke conditions

Speech dysarthria is a key symptom of neurological conditions like ALS, yet existing AI models designed to analyze it from audio signal rely on handcrafted features with limited inference performance. Deep learning approaches improve accuracy but lack interpretability. We propose an attention-based deep learning AI model to assess dysarthria severity based on listener effort ratings. Using 2,102 recordings from 125 participants, rated by three speech-language pathologists on a 100-point scale, we trained models directly from recordings collected remotely. Our best model achieved R2 of 0.92 and RMSE of 6.78. Attention-based interpretability identified key phonemes, such as vowel sounds influenced by ‘r’ (e.g., “car,” “more”), and isolated inspiration sounds as markers of speech deterioration. This model enhances precision in dysarthria assessment while maintaining clinical interpretability. By improving sensitivity to subtle speech changes, it offers a valuable tool for research and patient care in ALS and other neurological disorders.