Log-spectral amplitude estimation with Generalized Gamma distributions for speech enhancement

This paper presents a family of log-spectral amplitude (LSA) estimators for speech enhancement. Generalized Gamma distributed (GGD) priors are assumed for speech short-time spectral amplitudes (STSAs), providing mathematical flexibility in capturing the statistical behavior of speech. Although solutions are not obtainable in closed-form, estimators are expressed as limits, and can be efficiently approximated. When applied to the Noizeus database [8], proposed estimators are shown to provide improvements in segmental signal-to-noise ratio (SSNR) and COSH distance [14], relative to the LSA estimator proposed by Ephraim and Malah [2].

• Type: Journal
• Year: 2011
• Where: ICASSP
• Authors: Bengt J. Borgstrom, Abeer Alwan
• Keywords: Estimator, ICASSP 2011, Segmental Signal-to-noise Ratio, Short-time Spectral Amplitudes, Signal Processing