Presentation 1995/9/11
Speaker Identification using Dynamic Features of Speech and Feature Map
Tadashi Kitamura, Shinsai Takei,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We discuss for text-independent speaker recognition using dynamic features and a feature map. The model for each speaker consists of Kohonen's feature map and a predictive network. The first part of the model creates a feature map from the characteristic vectors of each speaker's speech and quantizes input speech into a specified vector on the map. The second part predicts transitional patterns of outputs on the feature map of the model and learns using a back-propagation algorithm. The final decision is made by a combination of VQ distortion and a prediction error. According to the speaker identification experiments for 8 speakers, it has been found that the combination of the feature map of mel cepstrum and the predictive network for the two-dimensional mel cepstrum gives the high score and exhibits the robustness for the time interval.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Speaker Identification / Feature Map / Predictive Network / Text Independent
Paper #
Date of Issue

Conference Information
Committee DSP
Conference Date 1995/9/11(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Digital Signal Processing (DSP)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Speaker Identification using Dynamic Features of Speech and Feature Map
Sub Title (in English)
Keyword(1) Speaker Identification
Keyword(2) Feature Map
Keyword(3) Predictive Network
Keyword(4) Text Independent
1st Author's Name Tadashi Kitamura
1st Author's Affiliation Nagoya Institute of Technology()
2nd Author's Name Shinsai Takei
2nd Author's Affiliation Nagoya Institute of Technology
Date 1995/9/11
Paper #
Volume (vol) vol.95
Number (no) 226
Page pp.pp.-
#Pages 7
Date of Issue