Noise reduction by noise-adaptive spectral magnitude expansion

Noise-adaptive spectral magnitude expansion bridges the gap between compandors used to overcome the limited dynamic range of a transmission channel and noise suppression techniques such as spectral subtraction or spectral short-time Wiener filtering. The underlying idea of noise-adaptive spectral magnitude expansion is to adapt the crossover point of the spectral magnitude expansion in each frequency channel based on the noise level. When no restrictions are imposed on the expansion characteristic, many types of spectral short-time filtering, such as spectral subtraction and spectral short-time Wiener filtering, can be considered as special cases of noise-adaptive spectral magnitude expansion.