Applying feature reduction analysis to a PPRLM-multiple Gaussian language identification system
作者： Luis Fernando D'haro Enríquez , Ricardo de Córdoba Herralde , Juan Manuel Lucas Cuesta
DOI:
关键词: Classifier (UML) 、 Language identification 、 Speech recognition 、 Dimensionality reduction 、 Missing data 、 Feature vector 、 Artificial intelligence 、 Gaussian 、 Pattern recognition 、 Mathematics 、 Feature selection 、 Utterance
摘要: This paper presents the application of a feature selection technique such as LDA to language identification (LID) system. The baseline system consists PPRLM module followed by multiple-Gaussian classifier. classifier makes use acoustic scores and duration features each input utterance. We applied dimension reduction space in order achieve faster easier-trainable imputed missing values our vectors before projecting them on new space. Our experiments show very low performance due approach. Using single projection error rates we have obtained are about 8.73% taking into account 22 most significant features.
参考文章(11)
Eliathamby Ambikairajah, Eric H. C. Choi, Liang Wang, Multi-layer kohonen self-organizing feature map for language identification. conference of the international speech communication association. pp. 174- 177 ,(2007)
Eliathamby Ambikairajah, Bo Yin, Fang Chen, Hierarchical language identification based on automatic language clustering. conference of the international speech communication association. pp. 178- 181 ,(2007)
Haizhou Li, Khe Chai Sim, Fusion of contrastive acoustic models for parallel phonotactic spoken language identification. conference of the international speech communication association. pp. 170- 173 ,(2007)
Qu Dan, Wang Bingxi, Mang Qiang, Two discriminative training schemes of GMM for language identification international conference on signal processing. ,vol. 1, pp. 630- 633 ,(2004) , 10.1109/ICOSP.2004.1452742
Edgar Acuña, Caroline Rodriguez, The Treatment of Missing Values and its Effect on Classifier Accuracy Springer, Berlin, Heidelberg. pp. 639- 647 ,(2004) , 10.1007/978-3-642-17103-1_60
R. Cordoba, R. San-Segundo, J. Macias, Juan Montero, R. Barra, L.F. D'Haro, J.C. Plaza, J. Ferreiros, Integration of acoustic information and PPRLM scores in a multiple-Gaussian classifier for Language Identification 2006 IEEE Odyssey - The Speaker and Language Recognition Workshop. pp. 1- 8 ,(2006) , 10.1109/ODYSSEY.2006.248105
Luis Fernando D'Haro, Fernando Fernández-Martínez, Juan Manuel Montero, Ricardo de Córdoba, Roberto Barra-Chicote, Language Identification using several sources of information with a multiple-Gaussian classifier conference of the international speech communication association. pp. 2137- 2140 ,(2007)
David K. Y. Chiu, BOOK REVIEW: "PATTERN CLASSIFICATION", R. O. DUDA, P. E. HART and D. G. STORK, Second Edition International Journal of Computational Intelligence and Applications. ,vol. 01, pp. 335- 339 ,(2001) , 10.1142/S1469026801000251
J. Braun, H. Levkowitz, Automatic language identification with recurrent neural networks international joint conference on neural network. ,vol. 3, pp. 2184- 2189 ,(1998) , 10.1109/IJCNN.1998.687199
R. San-Segundo, R. Córdoba, F. Fernández, J. Macías-Guarasa, A multiple-Gaussian classifier for Language Identification using acoustic information and PPRLM scores ,(2006)