Title :
Statistical linear approximation for environment compensation
Author_Institution :
Human & Comput. Interaction Lab., Samsung Adv. Inst. of Technol., Suwon, South Korea
Abstract :
The statistical linear approximation (SLA) method is proposed as a novel way to approximate a nonlinear function by a linearized model. In the proposed method, an optimization criterion for approximation is defined in terms of statistical expectation. The SLA is applied to environment compensation where the speech contamination rule appears as a highly nonlinear function of the relevant variables.
Keywords :
Gaussian distribution; approximation theory; functional analysis; least mean squares methods; optimisation; series (mathematics); speech processing; speech recognition; HMM; MSE; Taylor series expansion; environment compensation; experiments; linearized model; mean squared error; nonlinear function; nonlinear function approximation; optimization criterion; probability density functions; speaker-independent continuous digit recognition; speech contamination rule; speech recognition; statistical expectation; statistical linear approximation; variables; Contamination; Hidden Markov models; Linear approximation; Mathematical model; Optimization methods; Parameter estimation; Robustness; Speech recognition; Taylor series; Vectors;
Journal_Title :
Signal Processing Letters, IEEE
