A parametric formulation of the generalized spectral subtraction method

• Published in: IEEE transactions on speech and audio processing 1998-07, Vol.6 (4), p.328-337
• Main Authors: Boh Lim Sim, Yit Chow Tong, Chang, J.S., Chin Tuan Tan
• Format: Article
• Language: English
• Subjects: Acoustic noise; Amplitude estimation; Applied sciences; Automatic speech recognition; Constraint optimization; Exact sciences and technology; Gaussian noise; Information, signal and communications theory; Noise level; Noise reduction; Signal processing; Speech enhancement; Speech processing; Telecommunications and information theory; Testing
• ISSN: 1063-6676; 1558-2353
• DOI: 10.1109/89.701361

In this paper, two short-time spectral amplitude estimators of the speech signal are derived based on a parametric formulation of the original generalized spectral subtraction method. The objective is to improve the noise suppression performance of the original method while maintaining its computational simplicity. The proposed parametric formulation describes the original method and several of its modifications. Based on the formulation, the speech spectral amplitude estimator is derived and optimized by minimizing the mean-square error (MSE) of the speech spectrum. With a constraint imposed on the parameters inherent in the formulation, a second estimator is also derived and optimized. The two estimators are different from those derived in most modified spectral subtraction methods, which are predominantly nonstatistical. When tested under stationary white Gaussian noise and semistationary Jeep noise, they showed improved noise suppression results.