A complex-valued speech analysis using analytic signal realizes more accurate spectral estimation in low frequencies. We have already developed three kinds of time-varying complex AR (TV-CAR) parameter estimation algorithms for analytic speech signal, which are based on minimizing mean square error (MMSE), Huber's robust M-estimation and Instrumental Variable (IV) method. Needless to say, speech signal provides anti-resonance, especially in nasal speech. Accordingly, an ARMA parameter estimation may be useful on speech processing and speech research tool. This paper presents time-varying complex ARMA parameter estimation algorithm on the basis of the IV method, in which the excitation is assumed as the pulse train estimated from residual signal. The experiments with natural speech demonstrate that the proposed method achieves robust ARMA spectral estimation.
Index Terms. speech analysis, analytic signal, ARMA model, time-varying model, complex signal processing
Cite as: Funaki, K. (2001) A time-varying complex ARMA speech modeling based on IV method. Proc. Models and Analysis of Vocal Emissions for Biomedical Applications (MAVEBA 2001), 260-263
@inproceedings{funaki01_maveba,
author={Keiichi Funaki},
title={{A time-varying complex ARMA speech modeling based on IV method}},
year=2001,
booktitle={Proc. Models and Analysis of Vocal Emissions for Biomedical Applications (MAVEBA 2001)},
pages={260--263}
}