Feature extraction and recognition of Japanease voiceless stop consonants by principal component analysis

スポンサーリンク

概要

• 論文の詳細を見る

• At the present time, how to extract acoustic features of voiceless stop consonants is one of the most difficult probrems remaining unsolved in the field of automatic speech recognition. This paper describes a study on characteristic features of the Japanease voiceless stop consonants /p/, /t/ and /k/. A multi-dimensional statistical analysis method is applied to analyze their spectra. Analysis reveals that the principal characteristics discriminating between them are reflected in the accumulated power from about 1kHz up to 5kHz and the existence of a srectra peak. It is also found that these feature parameters of automatic recognition based on the maximum likelihood method are also performed. They are carried out on condition that the following vowel is correctly known beforehand and that there are no errors in detection of the noise onset. It is found that a recognition rate as high as about 97% can be attained for the training data set of known utterances if the recognition algorithm design to make the best use of the transition patterns of the feature parameters is adopted. It is also shown that the accurate detection of the moment of noise burst is essential to attain a high recognition rate.

• 社団法人日本音響学会の論文

著者

• Kobatake Hidefumi

  Department Of Electronic Engineering Faculty Of Technology Tokyo University Of Agriculture And Techn

• Noso Kazunori

  Electronic Laboratory,Central Engineering Laboratories,Nissan Motor Co,.Ltd.

• Noso Kazunori

  Electronic Laboratory Central Engineering Laboratories Nissan Motor Co .ltd.

関連論文

• Feature extraction and recognition of Japanease voiceless stop consonants by principal component analysis

スポンサーリンク