The use of speech recognition technology by people living with amyotrophic lateral sclerosis: a scoping review

Pareto-Optimized Non-Negative Matrix Factorization Approach to the Cleaning of Alaryngeal Speech Signals

The problem of cleaning impaired speech is crucial for various applications such as speech recognition, telecommunication, and assistive technologies. In this paper, we propose a novel approach that combines Pareto-optimized deep learning with non-negative matrix factorization (NMF) to effectively reduce noise in impaired speech signals while preserving the quality of the desired speech. Our method begins by calculating the spectrogram of a noisy voice clip and extracting frequency statistics. A threshold is then determined based on the desired noise sensitivity, and a noise-to-signal mask is computed. This mask is smoothed to avoid abrupt transitions in noise levels, and the modified spectrogram is obtained by applying the smoothed mask to the signal spectrogram. We then employ a Pareto-optimized NMF to decompose the modified spectrogram into basis functions and corresponding weights, which are used to reconstruct the clean speech spectrogram. The final noise-reduced waveform is obtained by inverting the clean speech spectrogram. Our proposed method achieves a balance between various objectives, such as noise suppression, speech quality preservation, and computational efficiency, by leveraging Pareto optimization in the deep learning model. The experimental results demonstrate the effectiveness of our approach in cleaning alaryngeal speech signals, making it a promising solution for various real-world applications.

Automatic Assessment of Aphasic Speech Sensed by Audio Sensors for Classification into Aphasia Severity Levels to Recommend Speech Therapies

Aphasia is a type of speech disorder that can cause speech defects in a person. Identifying the severity level of the aphasia patient is critical for the rehabilitation process. In this research, we identify ten aphasia severity levels motivated by specific speech therapies based on the presence or absence of identified characteristics in aphasic speech in order to give more specific treatment to the patient. In the aphasia severity level classification process, we experiment on different speech feature extraction techniques, lengths of input audio samples, and machine learning classifiers toward classification performance. Aphasic speech is required to be sensed by an audio sensor and then recorded and divided into audio frames and passed through an audio feature extractor before feeding into the machine learning classifier. According to the results, the mel frequency cepstral coefficient (MFCC) is the most suitable audio feature extraction method for the aphasic speech level classification process, as it outperformed the classification performance of all mel-spectrogram, chroma, and zero crossing rates by a large margin. Furthermore, the classification performance is higher when 20 s audio samples are used compared with 10 s chunks, even though the performance gap is narrow. Finally, the deep neural network approach resulted in the best classification performance, which was slightly better than both K-nearest neighbor (KNN) and random forest classifiers, and it was significantly better than decision tree algorithms. Therefore, the study shows that aphasia level classification can be completed with accuracy, precision, recall, and F1-score values of 0.99 using MFCC for 20 s audio samples using the deep neural network approach in order to recommend corresponding speech therapy for the identified level. A web application was developed for English-speaking aphasia patients to self-diagnose the severity level and engage in speech therapies.

Comparison of Slow and Forced Vital Capacity on Ability to Evaluate Respiratory Function in Bulbar-Involved Amyotrophic Lateral Sclerosis

Background and Objective

The percent-predicted forced vital capacity (FVC%) in the pulmonary function test (PFT) is generally used to evaluate the respiratory function in amyotrophic lateral sclerosis (ALS). The slow vital capacity (SVC) is another method to evaluate the respiratory function. Some neurologists found that the FVC% was not reflective of respiratory symptoms and the percent-predicted SVC (SVC%) was found to be higher in some patients with bulbar-onset ALS. We aimed to compare the percent predicted SVC (SVC%) with FVC% in evaluating the respiratory function and investigate the associations between the associations between clinical characteristics and the difference between the SVC% and the FVC% (SVC%-FVC%) in bulbar-involved ALS patients.

Method

This prospective study included patients with bulbar-involved ALS who visited the Peking University Third Hospital between October 2020 and November 2021. They underwent comprehensive clinical assessments, including bulbar symptom assessments, revised ALS functional rating scale (ALSFRS-R), Rasch-Built Overall Amyotrophic Lateral Sclerosis Disability Scale (Roads), and PFTs. The group differences were analyzed using parametric and non-parametric tests.

Results

A total of 59 participants were initially enrolled, and 51 of them were included in the final analysis. In patients with bulbar-involved ALS, the SVC% (73.82 ± 21.95%) was significantly higher (p = 0.013) than the FVC% (71.42 ± 23.15%). After controlling for other relevant variables, a partial correlation analysis showed a significant correlation (r = −0.352, p = 0.041) between ALSFRS-R1 score and SVC%-FVC%.

Conclusion

Our prospective study found that the SVC% was significantly higher and more reflective of actual respiratory function than the FVC% in patients with bulbar-involved ALS. Furthermore, the severity of dysarthria was found to be positively correlated with SVC%-FVC%, providing a clinical marker for predicting SVC%-FVC%.

Anthropomorphizing Technology: A Conceptual Review of Anthropomorphism Research and How it Relates to Children's Engagements with Digital Voice Assistants

‘Anthropomorphism’ is a popular term in the literature on human-technology engagements, in general, and child-technology engagements, in particular. But what does it really mean to ‘anthropomorphize’ something in today’s world? This conceptual review article, addressed to researchers interested in anthropomorphism and adjacent areas, reviews contemporary anthropomorphism research, and it offers a critical perspective on how anthropomorphism research relates to today’s children who grow up amid increasingly intelligent and omnipresent technologies, particularly digital voice assistants (e.g., Alexa, Google Assistant, Siri). First, the article reviews a comprehensive body of quantitative as well as qualitative anthropomorphism research and considers it within three different research perspectives: descriptive, normative and explanatory. Following a brief excursus on philosophical pragmatism, the article then discusses each research perspective from a pragmatistic viewpoint, with a special emphasis on child-technology and child-voice-assistant engagements, and it also challenges some popular notions in the literature. These notions include descriptive ‘as if’ parallels (e.g., child behaves ‘as if’ Alexa was a friend), or normative assumptions that human-human engagements are generally superior to human-technology engagements. Instead, the article reviews different examples from the literature suggesting the nature of anthropomorphism may change as humans’ experiential understandings of humanness change, and this may particularly apply to today’s children as their social cognition develops in interaction with technological entities which are increasingly characterized by unprecedented combinations of human and non-human qualities.