This paper presents the design of a companding non-uniform optimal scalar quantizer for speech coding. The quantizer is designed using two neural networks to perform the nonlinear transformation. These neural networks are used in the front and back ends of a uniform quantizer. Two approaches are presented in this paper namely adaptive critic designs (ACD) and particle swarm optimization (PSO), aiming to maximize the signal to noise ratio (SNR). The comparison of these optimal quantizer designs over bit rate range of 3 to 6 is presented. The perceptual quality of the coding is evaluated by the International Telecommunication Union''s Perceptual Evaluation of Speech Quality (PESQ) standard.
W. Zha and G. K. Venayagamoorthy, "Comparison of Non-Uniform Optimal Quantizer Designs for Speech Coding with Adaptive Critics and Particle Swarm," Conference Record of the 40th IAS Annual Meeting of the Industry Applications Conference, 2005, Institute of Electrical and Electronics Engineers (IEEE), Jan 2005.
The definitive version is available at http://dx.doi.org/10.1109/IAS.2005.1518380
40th IAS Annual Meeting of the Industry Applications Conference, 2005
Electrical and Computer Engineering
Keywords and Phrases
International Telecommunication Unions Perceptual Evaluation; PSO; SNR; Speech Quality Standard; Adaptive Codes; Adaptive Critic Design; Hearing; Neural Nets; Neural Network; Nonlinear Transformation; Nonuniform Optimal Scalar Quantizer; Particle Swarm Optimisation; Particle Swarm Optimization; Perceptual Quality; Quantisation (Signal); Signal to Noise Ratio; Speech Coding; Transform Coding
International Standard Serial Number (ISSN)
Article - Conference proceedings
© 2005 Institute of Electrical and Electronics Engineers (IEEE), All rights reserved.