Wavelet解析と線形予測法による音声特徴量抽出に関する研究

概要

論文の詳細を見る
Recently, the significance of smooth communication between human and computer in various fields has been increased The several methods of human speech analysis such as cepstrum analysis, Linear Predictive Analysis etc., have been suggested and evolved up to now. They are mainly related with the analysis-synthesis coding of speech signals in order to realize the characteristic parameters extraction of high quality in the low bit rate coding. On the other hand, the specification of speaker and the perception of emotion also come to be significant in IT technologies. However, there are many unsolved problems in the speech recognition methods, which are related specification of the speaker and to recognize their emotion. Hence, this study introduces the Wavelet Analysis which is taken notice of as an unsteady state signal processing method of Time-Frequency analysis. The purpose of this study is to combine Wavelet Analysis into the traditional speech processing methods, to suggest a new recognition method for the specification of speaker and the perception of emotion which may supply a lot of applications for IT technologies, and to find how to extract the characteristic parameters for the speakers. This study reveals that even though base frequency chnges in some ranges, the Fl and F2 does not change too much, and have an approximately constant value every individual. Therefore, individual difference distinction can be realized by position relations of formant of five vowels without being affected by emotions of the speaker.
沼津工業高等専門学校の論文
2008-01-31

Wavelet解析と線形予測法による音声特徴量抽出に関する研究

スポンサーリンク

概要

著者

関連論文

スポンサーリンク