id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_eyrgzf5g7neibnpi4o2sfe6bii	Leandro D. Vignolo	Genetic wavelet packets for speech recognition	2013	26	.pdf	application/pdf	11290	3334	81	genetic algorithm is able to find a representation that improves speech recognition results. make phoneme key-features more evident, in order to obtain significant improvements in the classification results [1]. used for speech recognition is built from the mel-frequency cepstral coefficients (MFCC) [3], which are based on a linear model of voice production extraction of high quality features from continuous wavelet coefficients according to signal classification criteria was presented. based on the best basis wavelet packet entropy method was proposed for electroencephalogram classification. approaches have been proposed for the optimization of wavelet based representations using swarm intelligence [25, 26]. Therefore, the fitness function was defined as a phoneme classifier, based on the optimized learning vector quantization (O-LVQ) techprocess of filtering and decimation was performed to obtain six decomposition levels, obtaining a full wavelet packet tree consisting of 1792 coefficients. our optimized representation is also based on wavelets, and the same problem could be expected, no post-processing was necessary for GWP.	./cache/work_eyrgzf5g7neibnpi4o2sfe6bii.pdf	./txt/work_eyrgzf5g7neibnpi4o2sfe6bii.txt