Design and Implementation of Optimal Noise Shaping Filters
When re-quantize a signal of low levels, serious distortion may occur due to word length reduction. Adding proper dither before re-quantization can improve the quantization error characteristic, while it will increase overall noise. By applying noise shaping, transposing most of the noise power to the band which human ear is less sensitive can reduce audible noise and thus improve auditory sense. This paper firstly discussed the applications of dither and noise shaping in professional digital audio, according to human ear’s psychoacoustic properties, by adopting genetic algorithm, fulfilled the design of psycho-acoustically optimal noise shaping filter. Then it applied the noise shaping technology to upsampling processing, realized the optimal noise shaping filter design under oversampling condition. Finally, it performed algorithm simulation on mat lab platform. Actual listening test confirms that by combining noise shaping technology with upsampling technology, sound quality can be improved significantly.
Chunliang Zhang and Paul P. Lin
J. L. Zhou and Z. Y. Zhu, "Design and Implementation of Optimal Noise Shaping Filters", Applied Mechanics and Materials, Vols. 226-228, pp. 2163-2166, 2012