Document Detail

Myoelectric signal classification for phoneme-based speech recognition.
MedLine Citation:
PMID:  17405376     Owner:  NLM     Status:  MEDLINE    
Traditional acoustic speech recognition accuracies have been shown to deteriorate in highly noisy environments. A secondary information source is exploited using surface myoelectric signals (MES) collected from facial articulatory muscles during speech. Words are classified at the phoneme level using a hidden Markov model (HMM) classifier. Acoustic and MES data was collected while the words "zero" through "nine" were spoken. An acoustic expert classified the 18 formative phonemes in low noise levels [signal-to-noise ratio (SNR) of 17.5 dB] with an accuracy of 99%, but deteriorated to approximately 38% under simulations with SNR approaching 0 dB. A fused acoustic-myoelectric multiexpert system, without knowledge of SNR, improved on acoustic classification results at all noise levels. A multiexpert system, incorporating SNR information, obtained accuracies of 99% at low noise levels while maintaining accuracies above 94% during low SNR (0 dB) simulations. Results improve on previous full word MES speech recognition accuracies by almost 10%.
Erik J Scheme; Bernard Hudgins; Phillip A Parker
Related Documents :
21090906 - Individual differences in working memory capacity predict retrieval-induced forgetting.
21387116 - The effect of listenability factors on the comprehension of police cautions.
4061286 - Combined effects of heat and noise on human performance: a review.
2356726 - Prediction of intelligibility of non-linearly processed speech.
17804576 - Different learned coordinate frames for planning trajectories and final positions in re...
20079856 - Inter-individual differences in resting-state functional connectivity predict task-indu...
Publication Detail:
Type:  Evaluation Studies; Journal Article; Research Support, Non-U.S. Gov't; Validation Studies    
Journal Detail:
Title:  IEEE transactions on bio-medical engineering     Volume:  54     ISSN:  0018-9294     ISO Abbreviation:  IEEE Trans Biomed Eng     Publication Date:  2007 Apr 
Date Detail:
Created Date:  2007-04-04     Completed Date:  2007-04-24     Revised Date:  2009-11-11    
Medline Journal Info:
Nlm Unique ID:  0012737     Medline TA:  IEEE Trans Biomed Eng     Country:  United States    
Other Details:
Languages:  eng     Pagination:  694-9     Citation Subset:  IM    
Institute of Biomedical Engineering, University of New Brunswick, Fredericton, NB E3B 5A3, Canada.
Export Citation:
APA/MLA Format     Download EndNote     Download BibTex
MeSH Terms
Electromyography / methods*
Expert Systems
Facial Muscles / physiology*
Muscle Contraction / physiology*
Pattern Recognition, Automated / methods*
Reproducibility of Results
Sensitivity and Specificity
Speech / physiology*
Speech Production Measurement / methods*
Speech Recognition Software*

From MEDLINE®/PubMed®, a database of the U.S. National Library of Medicine

Previous Document:  Estimation of elbow-induced wrist force with EMG signals using fast orthogonal search.
Next Document:  A reconstruction algorithm for breast cancer imaging with electrical impedance tomography in mammogr...