Experiment of Masking-Based β-Order MMSE Speech Enhancement in a Car Noise Environment

In this paper, we investigate the use of masking-based β-order MMSE (MβOM) speech enhancement algorithm as applied to a car noise environment. Based on the previous work on the β-order MMSE speech enhancement, it is found that MβOM has a satisfactory performance in the case of speech corrupted by Volvo interior car noise from the NOISEX-92 database. In this paper, the adaptation of the original MβOM is modified according to the characteristics of the car noise. The particular type of the noise of a car moving in the highway is considered. The experimental results are shown in terms of both Itakura Saito (IS) distortion and modified Bark spectral distortion (MBSD) measures.