Abstract | ||
---|---|---|
We present a new unified training scheme using a feature extractor and HMM classifiers for better speech recognition performance. Both feature extractor and classifier are trained simultaneously to minimize classification error. Multiframe features are extracted using spectro-temporal dynamics and the feature extractor is implemented as a multilayer network, which is trained by a backpropagation (BP) algorithm with the help of an HMM inversion algorithm. The initial parameter values of the feature extractor are set for Mel-frequency cepstral coefficients (MFCC) as well as their delta and acceleration components. The experiments for phoneme classification demonstrate the practicality of unified training. |
Year | DOI | Venue |
---|---|---|
2012 | 10.1109/LSP.2011.2179647 | IEEE Signal Process. Lett. |
Keywords | Field | DocType |
mel-frequency cepstral coefficients,delta components,hmm inversion algorithm,speech recognition,feature extraction,classification error minimization,unified training scheme,backpropagation algorithm,hidden-markov model classifier,feature learning,feature extractor,backpropagation,spectro-temporal dynamics,multilayer network,bp algorithm,phoneme classification,acceleration components,unified feature extractor and classifier,hidden markov models,mfcc,hmm classifier,discrete cosine transform,hidden markov model,mel frequency cepstral coefficient | Mel-frequency cepstrum,Pattern recognition,Computer science,Feature extraction,Speech recognition,Artificial intelligence,Extractor,Acceleration,Speech recognition performance,Hidden Markov model,Classifier (linguistics),Backpropagation | Journal |
Volume | Issue | ISSN |
19 | 2 | 1070-9908 |
Citations | PageRank | References |
0 | 0.34 | 9 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Jung-Hui Im | 1 | 0 | 0.34 |
Soo-Young Lee | 2 | 1137 | 163.87 |