Nonlinear spectral subtraction method for colored noise reduction using multi-band Bark scale

We propose an improved spectral subtraction method for the reduction of colored acoustic noise added to the speech in a helicopter cockpit or in a car caused by the engine sound. Our implementation uses an over-estimation subtraction method for spectral subtraction. The noise signal does not affect uniformly the speech signal over the whole spectrum. A nonlinear multi-band Bark scale frequency spacing approach is proposed to reduce the colored noise. Simulations show a better quality in terms of Itakura-Saito distance and perceptual evaluation of quality for the enhanced speech.