Noise Reduction by Maximum a Posteriori Spectral Amplitude Estimation with Supergaussian Speech Modeling

ESTIMATION WITH SUPERGAUSSIAN SPEECH MODELING Thomas Lotter and Peter Vary Institute of Communication Systems and Data Processing ( ) Aachen University (RWTH), Templergraben 55, D-52056 Aachen, Germany E-mail: lotter vary @ind.rwth-aachen.de ABSTRACT This contribution presents a spectral amplitude estimator for acoustical background noise suppression based on maximum a posteriori estimation and supergaussian statistical modeling of the speech DFT coefficients. The probability density function of the speech spectral amplitude is modeled with a simple parametric function, which allows a high approximation accuracy for Laplace or Gamma distributed real and imaginary parts of the speech DFT coefficients. Based on the approximation, a computationally efficient maximum a posteriori speech estimator is derived, which outperforms the Ephraim-Malah algorithm in a single channel noise reduction framework.