Resumo -Este artigo apresenta uma nova tecnica de projeto de dicionarios de vetores esparsos para codificadores CELP (Code Excited Linear Prediction), baseada no esquema de ex citacao multipulso e no algoritmo LBG (Linde-Buzo-Gray) de treinamento. 0 dicionario resultante -denorninado multi pulso ou, simplesmente, :MP-possui as vantagens inerentes aos dicionarios esparsos convencionais. Seu grande merito e ter ainda a vantagem de poder ser otimizado atraves de urn procedimento de treinamento. Atraves de simulacoes, 0 di cionario :MP foi comparado com os dicionarios gaussiano e "vector-sum" (VS). Os resultados mostraram que, dentre os tres tipos de dicionario testados, 0 dicionario :MP e0 que tern melhor desempenho em termos da qualidade da voz codifi cada. Neste artigo, sao apresentadas tambem comparacoes quanta acomplexidade associada aos tres dicionarios, Abstract -This paper presents a new algorithm to design co debooks of sparse vectors for code excited linear prediction (CELP) speech encoders. The algorithm is based on multi pulse excitations and on the LBG codebook training algo rithm. The codebooks, which are referred to as multi pulse (:MP) codebooks, have the inherent advantages of sparse co debooks. Moreover, they can be optimized by a closed-loop training procedure. We present a comparative analysis of the:MP codebook with vector-sum (VS) and Gaussian code books. Simulation results show that the :MP codebook yields the best performance in terms of speech quality. Complexities of these codebooks are also compared.Palavras chaves -Codificacao de Voz, CELP, Otimizacao de Dicionario.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.