Presentation | 1996/9/13 Reduction of LPC Spectrum Dimension Using a Wine-Glass-Type Neural Network Hironori ITO, Shoji KAJITA, Kazuya TAKEDA, Fumitada ITAKURA, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Reducing the dimension of acoustic feature space is realized using a wine-glass-type neural network, which has the fewer number units in middle layer than the input and output layers, trained for the identity mapping. A wine-glass-type neural network, which has 32 units for both input and output layers and two to five units for the middle layer are trained so as to map the input of 32 dimensional LPC spectrum to the identical output vectors. After neural network is trained, signal to deviation ratio (SDR) of log spectrum is smaller than using KL expansion. Moreover, DTW isolated word recognition experiments are performed using 123 similar city name utterances of a male speaker. Using the output of the middle layer units reduced to 3-5 feature vector, the recognition accuracy are higher than using KL expansion. Therefore the effectiveness of nonlinear identity mapping using neural network for reducing the feature dimension is confirmed. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | wine-glass-type neural network / LPC spectrum / identity mapping / dimension reduction / word recognition |
Paper # | DSP-96-79,SP-96-54 |
Date of Issue |
Conference Information | |
Committee | DSP |
---|---|
Conference Date | 1996/9/13(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) | Reduction of LPC Spectrum Dimension Using a Wine-Glass-Type Neural Network |
Sub Title (in English) | |
Keyword(1) | wine-glass-type neural network |
Keyword(2) | LPC spectrum |
Keyword(3) | identity mapping |
Keyword(4) | dimension reduction |
Keyword(5) | word recognition |
1st Author's Name | Hironori ITO |
1st Author's Affiliation | Graduate School of Engineering, Nagoya University() |
2nd Author's Name | Shoji KAJITA |
2nd Author's Affiliation | Graduate School of Engineering, Nagoya University |
3rd Author's Name | Kazuya TAKEDA |
3rd Author's Affiliation | Graduate School of Engineering, Nagoya University |
4th Author's Name | Fumitada ITAKURA |
4th Author's Affiliation | Graduate School of Engineering, Nagoya University |
Date | 1996/9/13 |
Paper # | DSP-96-79,SP-96-54 |
Volume (vol) | vol.96 |
Number (no) | 239 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |