TOPICS IN STATISTICAL QUANTIZATIONS.
Abstract
The effects of the application of amplitude quantization to stochastic signals are analyzed and discussed. First- and second-order properties of the quantized signal and of the quantization noise are developed. Considerations for the optimum quantization of signals are given, with examples using speech signals and Gaussian signals. The optimum compressor characteristic for Gaussian signals is derived for a minimum mean-square error criterion. A special case of the problem of signal detectability following a band-pass limiter is reviewed. The use of dither signals (supplementary signals added to the desired signal before quantization) in conjunction with amplitude quantization is discussed and analyzed for combinations of Gaussian and uniform-density signals. Measures of system performance are discussed, and optimum dither signal levels are derived. Experimental results are given in support of the analytical results. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1965
- Accession Number
- AD0471027
Entities
People
- Harry Nicholson Shaver
Organizations
- Stanford University