Speech Recognition Using Myoelectric Signals With Neural Networks

Morse, Michael S.; Gopalan, Y.N.; Wright, Matthew

doi:10.1109/iembs.1991.684800

Cited by 20 publications

(12 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, for vocabularies over 10 words, the recognition accuracy remained below 70% [254,257] until 2001 when Chan et al [253] achieved classification accuracies of 97% and 90% for two subjects on a 10-word vocabulary of English digits. They used five pairs of gel Ag/AgCl electrodes to measure sEMG signals from five facial muscles: levator anguli oris, zygomaticus major platysma, depressor anguli oris and anterior belly of the digastric.…”

Section: Silent Speech Recognitionmentioning

confidence: 96%

Current state of digital signal processing in myoelectric interfaces and related applications

Hakonen

Piitulainen

Visala

2015

Biomedical Signal Processing and Control

257

185

View full text Add to dashboard Cite

a b s t r a c tThis review discusses the critical issues and recommended practices from the perspective of myoelectric interfaces. The major benefits and challenges of myoelectric interfaces are evaluated. The article aims to fill gaps left by previous reviews and identify avenues for future research. Recommendations are given, for example, for electrode placement, sampling rate, segmentation, and classifiers. Four groups of applications where myoelectric interfaces have been adopted are identified: assistive technology, rehabilitation technology, input devices, and silent speech interfaces. The state-of-the-art applications in each of these groups are presented.

show abstract

Section: Silent Speech Recognitionmentioning

confidence: 96%

Current state of digital signal processing in myoelectric interfaces and related applications

Hakonen

Piitulainen

Visala

2015

Biomedical Signal Processing and Control

257

185

View full text Add to dashboard Cite

show abstract

“…The research also concludes that the measurement parameters like EMG and mechanical movements play a significant role in the outcome of experiment and, hence, need to be addressed in conjunction. As the future scope of work, the features from the crosscorrelation profile can be used to predict the unspoken words in silent speech recognition [5,6]. The investigation on the CC in JML co-ordination can be used in designing various biomedical engineering tools.…”

Section: Discussionmentioning

confidence: 99%

“…It outperforms the speech perception in a noisy environment [1]. The suitable feature extraction leads to easy control of prosthetic voice after Laryngectomy [2], emotion identification [3,4], silent speech recognition [5,6] and oral rehabilitation [7,8]. In all these important applications the facial muscles co-ordination plays a significant role.…”

Section: Introductionmentioning

confidence: 99%

Cross-correlation evaluated muscle co-ordination for speech production

Joshi

Bhatia

2013

Journal of Medical Engineering & Technology

View full text Add to dashboard Cite

In this investigation the jaw-mouth-lip (JML) co-ordination is quantified using cross-correlation (CC) between facial electromyography (EMG) while speaking words 'zero to nine'. The facial EMG data of four prominent facial muscles, namely Zygomaticus Major (ZM) from the mouth region, Levator Labii Superioris (LLS), Orbicularis Oris Inferioris (OOI) from the lip region and Anterior Belly of Diagastric (ABD) from the jaw region are collected and analysed. A total of six muscle pairs, (1) ABD-ZM, (2) ZM-LLS, (3) OOI-ZM, (4) ABD-LLS, (5) ABD-OOI and (6) LLS-OOI, are formed based on their position. Based on correlation study, all these muscle pairs show a significantly different co-ordination level (p<0.05). The CC of the ABD-LLS muscle pair is observed with maximum co-ordination level among all spoken words (Tukey's post-hoc, α=0.05). These investigations based on CC in JML co-ordination could be used in designing various biomedical engineering tools such as speech prosthesis, facial lie detectors, etc.

show abstract

“…EMG provides digital information of the physical muscles. EMG records speech perception in a noisy environment [1], silent speech recognition [2,3] and emotions identification [4,5], and a suitable feature extraction of facial EMG leads to easy control of prosthetic voice after laryngectomy [6]. Despite there being several studies on facial EMG, research into gender classification using facial EMG is yet to be nourished.…”

Section: Introductionmentioning

confidence: 99%

Feasibility of facial EMG in gender classification during speech production

Godiyal

Sharma

Joshi

et al. 2012

Journal of Medical Engineering & Technology

View full text Add to dashboard Cite

Gender identification has its unique importance in sports and forensic sciences. However, the social issues are major constraints to identify the gender of a human being. A novel approach for gender classification using facial EMG is presented. The time domain features of facial EMGs are explored towards their robustness in gender identification. A Davies-Bouldin index is calculated to evaluate the features. Anterior belly of digastrics along with Orbicularis Oris Inferior showed the promising classification accuracy of 80% with the reported features. A possible enhancement in accuracy could be possible by a larger dataset.

show abstract

Speech Recognition Using Myoelectric Signals With Neural Networks

Cited by 20 publications

References 4 publications

Current state of digital signal processing in myoelectric interfaces and related applications

Current state of digital signal processing in myoelectric interfaces and related applications

Cross-correlation evaluated muscle co-ordination for speech production

Feasibility of facial EMG in gender classification during speech production

Contact Info

Product

Resources

About