Wideband Audio Compression using Subband Coding and Entropy-Constrained Scalar Quantization.
Abstract
Source coding of wideband audio signals for storage applications and/or transmission over band limited channels is currently a research topic receiving considerable attention. A goal common to all systems designed for wideband audio coding is to achieve an efficient reduction in code rate, while maintaining imperceptible differences between the original and coded audio signals. In this thesis, an effective source coding scheme aimed at reducing the code rate to the entropy of the quantized audio source, while providing good subjective audio quality, is discussed. This scheme employs the technique of subband coding, where a 32-band single sideband modulated filter bank is used to perform subband analysis and synthesis operations. Encoding and decoding of the subbands is accomplished using entropy constrained uniform scalar quantization and subsequent arithmetic coding. A computationally efficient subband rate allocation procedure is used which relies on analytic models to describe the rate distortion characteristics of the subband quantizers. Signal quality is maintained by incorporating masking properties of the human ear into this rate allocation procedure. Results of simulations performed on compact disc quality audio segments are provided.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1995
- Accession Number
- ADA296328
Entities
People
- Trevor R. Trinkaus
Organizations
- Rome Laboratory