Adaptation of Acoustic Models in Joint Speaker and Noise Space Using Bilinear Models

Yongwon JEONG  Hyung Soon KIM  

IEICE TRANSACTIONS on Information and Systems   Vol.E97-D   No.8   pp.2195-2199
Publication Date: 2014/08/01
Online ISSN: 1745-1361
DOI: 10.1587/transinf.E97.D.2195
Type of Manuscript: LETTER
Category: Speech and Hearing
bilinear model,  eigenvoice speaker adaptation,  environment adaptation,  speaker adaptation,  speech recognition,  

Full Text: PDF(100.3KB)>>
Buy this Article

We present the adaptation of the acoustic models of hidden Markov models (HMMs) to the target speaker and noise environment using bilinear models. Acoustic models trained from various speakers and noise conditions are decomposed to build the bases that capture the interaction between the two factors. The model for the target speaker and noise is represented as a product of bases and two weight vectors. In experiments using the AURORA4 corpus, the bilinear model outperforms the linear model.