Linear-scale perceptual feature extraction for Speech Bandwidth Extensions

IEICE Electronics Express

Koeng-Mo Sung¹  Kuekjae Lee¹  Sang Bae Chon¹  Mingu Lee¹ 

[1] Applied Acoustics Lab., Institute of New Media and Communications, Department of Electrical Engineering, Seoul National University

关键词: BWE;    NMF;    MFCCs;   

DOI  :  10.1587/elex.8.1143

学科分类：电子、光学、磁材料

来源: Denshi Jouhou Tsuushin Gakkai

【 摘 要 】

References(6)This paper presents a new method to extract linear-scale perceptual feature as a subsitute of MFCCs for highband (3.4kHz∼) in Speech Bandwidth Extensions(BWE). The feature extraction method is based on the mel-scale constrained Nonnegative Matrix Factorization(NMF), which decompose linear-scale log spectrum into a linear combination of mel-scale latent variables. While MFCCs parametrization contains non-invertible procedures, suggested feature is represented in linear-scale and proper to recover the highband time-domain speech. Experiment results report that suggested feature shows better instrumental performance with narrowband MFCCs than real cepstrum without additional computation.

【 授权许可】

Unknown