Title :
Robust dynamic tone feature extraction using 2D oriented filters
Author_Institution :
Sch. of SEDC, Queensland Univ., Brisbane, Qld., Australia
fDate :
3/17/2005 12:00:00 AM
Abstract :
A novel method is proposed to extract dynamic tone features using 2D oriented filters. As 2D oriented filters are capable of capturing pitch orientations, they can make full use of the information from neighbouring frames to extract dynamic tone features without knowing the pitch or using pitch tracking algorithms. Experimental results show that the proposed method can produce dynamic tone features of good quality.
Keywords :
feature extraction; filtering theory; low-pass filters; speech processing; 2D oriented filters; dynamic tone feature extraction; lowpass filter; pitch orientation; pitch tracking algorithms; speech processing systems; speech recognition systems;
Journal_Title :
Electronics Letters
DOI :
10.1049/el:20057382