F-15 A Robust GMM Based Speaker Recognition System with Features Extracted from Voiced Parts

スポンサーリンク

概要

• 論文の詳細を見る

• A robust text independent speaker identifications system is proposed in which the speaker is characterized using the pitch period and LPC-CEPSTRAL extracted from voiced parts. These features are feed into a Gaussian Mixture Models used for the speaker identification. The proposed system was evaluated using a data-base of 80 speakers, with phrases of 3-5s in Japanese language stored during 4 months. Evaluation results show that proposed system achieves a global recognition rate of more than 95%.

• FIT(電子情報通信学会・情報処理学会)推進委員会の論文
• 2002-09-13

著者

• Perez-meana Hector

  Esime Culhuacan National Polytechnic Institute Of Mexico

• Acevedo Eric

  ESIME Culhuacan, National Polytechnic Institute of Mexico

• Nagai Takayuki

  The University of Electro Communications

• Kurematsu Akira

  The University of Electro Communications

• Nakano-Miyatake Mariko

  ESIME Culhuacan, National Polytechnic Institute of Mexico

• Nakano-miyatake Mariko

  Esime Culhuacan National Polytechnic Institute Of Mexico

• Acevedo Eric

  Esime Culhuacan National Polytechnic Institute Of Mexico

• Nagai T

  The University Of Electro Communications

• Kurematsu A

  The University Of Electro Communications

関連論文

• F-15 A Robust GMM Based Speaker Recognition System with Features Extracted from Voiced Parts
• D-14-16 Speaker Recognition using Gaussian Mixtures Model based on Combination of Spectral and Pitch Features
• Fast LOT with Unequal Length Basis Functions : Realization and Application in Subband Image Coding

スポンサーリンク