Over the last two decades, myoelectric signals have been widely used in fields including rehabilitation devices and human-machine interfaces. This study aimed to develop an algorithm for surface electromyography (sEMG) data acquisition utilizing low-cost hardware and validate its performance using English vowels as silent speech content. The sEMG data were collected from the three facial muscles of one healthy subject. The sEMG signals were pre-processed, and various time-domain and statistical features were extracted in real time. The raw data and features were then used to train and test three customized machine learning classifiers: k-nearest neighbor (KNN), support vector machine (SVM), and artificial neural network (ANN). All customized classifiers achieved almost equivalent accuracy rates of 0.83 ± 0.01 in recognizing the English vowels with an improvement of 27.27% (KNN), 3.75% (SVM), and 51.85% (ANN) utilizing the same low-cost data acquisition hardware. Our findings are substantially closers to the results of commercial hardware setups, which raise the possibility of potential usage of low-cost sEMG data acquisition systems with the proposed algorithm in place of commercial hardware setups for rehabilitation devices and other related sectors of human-machine interaction.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.