• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 1
  • Tagged with
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

Aplica??o do m?todo de fus?o para verifica??o de locutor independente de texto

Silva, Mayara Ferreira da 10 July 2015 (has links)
Submitted by Setor de Tratamento da Informa??o - BC/PUCRS (tede2@pucrs.br) on 2016-01-04T17:56:48Z No. of bitstreams: 1 DIS_MAYARA_FERREIRA_DA_SILVA_COMPLETO.pdf: 2803272 bytes, checksum: 9305b74451ec83ddca38d1c444ffb3dd (MD5) / Made available in DSpace on 2016-01-04T17:56:48Z (GMT). No. of bitstreams: 1 DIS_MAYARA_FERREIRA_DA_SILVA_COMPLETO.pdf: 2803272 bytes, checksum: 9305b74451ec83ddca38d1c444ffb3dd (MD5) Previous issue date: 2015-07-10 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior - CAPES / This work presents an overview of text independent speaker verification, describing the basic operation of the system and the reviewing some important developments in speaker modeling and feature extraction from speech. Following, a point of improvement identified within the feature extraction stage leads to the main objective of this work: to determine one or more sets of coefficients relevant to speaker discrimination while minimizing the equal error rate (EER). The proposal is to replace the delta(?) and double-delta(??) coefficients by a linear predictor code (LPC) for the mel frequency cepstral coefficients (MFCC). In addition, score level fusion is employed to combine the ouputs of MFCC-only and MFCC-LPC systems, as well as MFCC-only and MFCC-?-?? systems. In all cases, performance is evaluated with respect to variations of the signal to noise-ratio (SNR) in the tested audio. In addition, the work introduces a new Brazilian Portuguese speech repository containing free-speech from 155 males. Results and discussions are presented with a reflection on the expected outcomes, as well as general comments and observations. Finally, concludings remarks are made about the work, featuring future prospects regarding text independent speaker verification research. This work attained a 4% reduction in the EER compared to the reference system (MFCC-only), with best results occuring in the case fusion of MFCC-only and MFCC-?-?? scores. / Este trabalho apresenta uma vis?o geral acerca de verifica??o de locutor independente de texto, demonstrando o funcionamento b?sico do sistema e as principais refer?ncias de m?todos j? utilizados ao longo de anos para extra??o de caracter?sticas da fala e modelamento do locutor. Detectado um ponto a ser trabalhado dentro da etapa de extra??o de caracter?sticas, objetiva-se determinar coeficientes ou um conjunto destes relevantes para discrimina??o do locutor, com o intuito de minimizar a EER (Equal Error Rate). A proposta consiste em substituir os coeficientes delta(?) e double-delta(?2) por coeficientes de um preditor LPC (Linear Predictor Coding) o qual realiza a predi??o dos coeficientes MFCC (Mel Frequency Cepstral Coeficients). Al?m disso, aplica-se uma fus?o a n?vel de score em fun??o de sistemas baseados em MFCC e LPC. Outra an?lise discutida no trabalho ? a fus?o de um sistema MFCC com ? e ??. Um t?pico tamb?m avaliado ? com rela??o a varia??es de SNRs (Signal to Noise Ratios) nos ?udios testados. Al?m disso, ? elaborado um banco de falas em portugu?s brasileiro. Por fim, s?o expostos os resultados obtidos e ? feita a an?lise dos mesmos, a fim de refletir sobre o que era esperado e levantar alguns coment?rios. Enfim, s?o feitas as considera??es a respeito do trabalho, e elencadas as perspectivas futuras em torno das pesquisas de verifica??o de locutor independente de texto. Com este trabalho atingiu-se uma redu??o de 4% na taxa de erro igual (EER) em compara??o ao sistema de refer?ncia, sendo que os melhores resultados foram apresentados pelo sistema que realiza um fus?o do sistema MFCC com o ? e ??.

Page generated in 0.031 seconds