Presentation 1997/11/28
A Method for Noisy Speech Recognition Using Weighting Spectral Parameters
Kiyotaka NAGAO, Masanori AKITA, Naoshi ARIKAWA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This report shows the method for improving the result of speech recognition under the noisy environment. Former reports show that the spectral subtraction and the recovery of valleys in spectral envelope obtained from polluted speech are effective for the improvement of the recognition results. However, these methods are not enough for the recognition under the poor environment. And the rules of adding the valleys for polluted spectral envelopes are complicated and sensitive. In this report, the averaged spevtral envelope obtained from 5 vowel spectrum are used for the emphasis of valleys. The vocalic spectral information at lower frequency range is emphasized and the spectrum obtained from consonants is not changed. In this experiment, the emphasis coefficients are varied on cepstral domain. This method is used for the recognition of noisy digits and improvement of recognition results are shown.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Speech recognition / Colored noise / Mel-cepstrum / Spectral subtraction / Quefrency
Paper # EA97-70
Date of Issue

Conference Information
Committee EA
Conference Date 1997/11/28(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 Engineering Acoustics (EA)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Method for Noisy Speech Recognition Using Weighting Spectral Parameters
Sub Title (in English)
Keyword(1) Speech recognition
Keyword(2) Colored noise
Keyword(3) Mel-cepstrum
Keyword(4) Spectral subtraction
Keyword(5) Quefrency
1st Author's Name Kiyotaka NAGAO
1st Author's Affiliation Faculty of Engineering, Oita University()
2nd Author's Name Masanori AKITA
2nd Author's Affiliation Faculty of Engineering, Oita University
3rd Author's Name Naoshi ARIKAWA
3rd Author's Affiliation Faculty of Engineering, Oita University
Date 1997/11/28
Paper # EA97-70
Volume (vol) vol.97
Number (no) 406
Page pp.pp.-
#Pages 8
Date of Issue