Noise environment speech recognition using spectral subband centroids

15 References

Bibliographic Information

Other Title
  • スペクトルサブバンドセントロイドを用いた雑音下での音声認識

Search this article

Description

This paper investigates the effectiveness of a novel feature for speech recognition called spectral subband centroids (SSC). SSC are computed as frequency centroids for each subband using the power spectrum of the speech signal. This feature can be obtained reliably even under noisy conditions because SSC are mainly computed from spectral peaks such as formants whose positions are almost unchanged in a noisy environment. Therefore, we can expect SSC to provide here useful information. Experimental results on Japanese spontaneous speech recognition showed that SSC produced significant improvements at SNR=10dB and 20dB when used as a supplemental feature to the conventional Mel-Frequency Cepstral Coefficients (MFCC).

Journal

  • IPSJ SIG Notes

    IPSJ SIG Notes 19 23-28, 1997-12-11

    Information Processing Society of Japan (IPSJ)

References(15)*help

See more

Keywords

Details 詳細情報について

  • CRID
    1570009752305340928
  • NII Article ID
    110002954468
  • NII Book ID
    AN10442647
  • ISSN
    09196072
  • Text Lang
    ja
  • Data Source
    • CiNii Articles

Report a problem

Back to top