Paper Info
Title
A comparative study of noise estimation algorithms for nonlinear compensation in robust speech recognition.
Abstract
The Gauss-Newton and EM-FA methods are two main approaches for estimating noise parameters of the non-linear compensation models.We conduct a systematic comparison between these two estimation methods.Both approaches belong to the family of gradient-based methods except with different convergence rates.The Gauss-Newton method is superior to the EM-FA method in terms of convergence. Nonlinear compensation models make use of a nonlinear mismatch function, which characterizes the joint effects of additive and convolutional noise, to realize noise-robust speech recognition. Representative compensation models consist of vector Taylor series (VTS), data-driven parallel model combination (DPMC), and unscented transform (UT). The noise parameters of the compensation models, often estimated in the maximum likelihood (ML) sense, are known to play an important role on the system performance in noisy conditions. In this paper, we conduct a systematic comparison between two popular approaches for estimating the noise parameters. The first approach employs the Gauss-Newton method in a generalized EM framework to iteratively maximizing the EM auxiliary function. The second approach views the compensation models from a generative perspective, giving rise to an EM algorithm, analogous to the ML estimation for factor analysis (EM-FA). We demonstrate a close connection between these two approaches: they belong to the family of gradient-based methods except with different convergence rates. Note that the convergence property can be crucial to the noise estimation since model compensation may be frequently carried out in changing noisy environments for retaining desired performance. Furthermore, we present an in-depth discussion on the advantages and limitations of the two approaches, and illustrate how to extend these approaches to allow for adaptive training. The investigated noise estimation approaches are evaluated on several tasks. The first is to fit a GMM model to artificially corrupted samples, and then speech recognition are performed on the Aurora 2 and Aurora 4 tasks.
Year
DOI
Venue
2017
10.1016/j.specom.2017.02.005
Speech Communication
Keywords
Field
DocType
Factor analysis,Gauss-Newton method,Nonlinear compensation,Parallel model combination,Robust speech recognition,Vector Taylor series
Convergence (routing),Nonlinear system,Gauss newton method,Computer science,Maximum likelihood,Auxiliary function,Artificial intelligence,Taylor series,Pattern recognition,Expectation–maximization algorithm,Algorithm,Unscented transform,Speech recognition
Journal
Volume
Issue
ISSN
89
C
0167-6393
Citations 
PageRank 
References 
0
0.34
16
Authors
2
Name
Order
Citations
PageRank
Yong Zhao112713.62
Biing-Hwang Juang23388699.72