Speaker Adaptation Based on PPCA of Acoustic Models in a Two-Way Array Representation

Yongwon JEONG  

Publication
IEICE TRANSACTIONS on Information and Systems   Vol.E97-D   No.8   pp.2200-2204
Publication Date: 2014/08/01
Online ISSN: 1745-1361
DOI: 10.1587/transinf.E97.D.2200
Type of Manuscript: LETTER
Category: Speech and Hearing
Keyword: 
expectation-maximization,  probabilistic principal component analysis,  speaker adaptation,  speech recognition,  

Full Text: PDF(90.3KB)>>
Buy this Article




Summary: 
We propose a speaker adaptation method based on the probabilistic principal component analysis (PPCA) of acoustic models. We define a training matrix which is represented in a two-way array and decompose the training models by PPCA to construct bases. In the two-way array representation, each training model is represented as a matrix and the columns of each training matrix are treated as training vectors. We formulate the adaptation equation in the maximum a posteriori (MAP) framework using the bases and the prior.