Study of a Novel ADM Algorithm with Pre-processing for Performance Improvement

In this paper, a novel technique of a step-size adaptation algorithm for improving the signal-to-noise ratio (SNR) of Adaptive Delta Modulators (ADM) with one-bit quantization is presented. The two existing step-size algorithms--SONG and modified ABATE are discussed and also their performance comparison is made. For further improvement in the SNR with the proposed ADM includes an ADM with pre-processed signals which is also described. Large variations in the input signal require larger step sizes. If step size is not increased, it leads to slope-overload error. These large variations in the input signal are removed/smoothened by a suitable pre-processing method, one of which is using an integrator that can smoothen the rapid changes. From simulations it is confirmed that the new step-size algorithm for the ADM with pre-processing (proposed ADM) leads to an improvement in the SNR compared to the conventional ADM.

[1]  Jerry D. Gibson,et al.  Structures for SNR scalable speech coding , 2006, IEEE Transactions on Audio, Speech, and Language Processing.

[2]  G. S. Tombras,et al.  New adaptation algorithm for a two-digit adaptive delta modulation system , 1990 .

[3]  J.D. Gibson,et al.  Speech coding methods, standards, and applications , 2005, IEEE Circuits and Systems Magazine.

[4]  Chong Un,et al.  A Study of the Comparative Performance of Adaptive Delta Modulation Systems , 1980, IEEE Trans. Commun..

[5]  Mansour A. Aldajani,et al.  A stable adaptive structure for delta modulation with improved performance , 2001, 2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.01CH37221).

[6]  George S. Tombras,et al.  A 2-bit Adaptive Delta Modulation System with Improved Performance , 2007, EURASIP J. Adv. Signal Process..

[7]  Mansour A. Aldajani,et al.  Stability and performance analysis of an adaptive sigma-delta modulator , 2001 .

[8]  Ming Yang.  Low bit rate speech coding , 2004, IEEE Potentials.

[9]  Satoshi Nakamura,et al.  Noise adaptive speech recognition with acoustic models trained from noisy speech evaluated on Aurora-2 database , 2002, INTERSPEECH.