Trellis noise shaping converters architecture and evaluation

摘要

A new approach to improving performance of noise-shaping converters is introduced. Trellis Noise Shaping Converters, using trellis structure and Viterbi algorithm, enable data bits in 1-bit high-sampling audio format to be utilized more efficiently, almost optimally in information-theoretic sense, which yields better performance in stability, signal to noise ratio and tonal behavior. Its architecture and performance are briefly examined. Based on the unsatisfactory performance of conventional converters, several leading researchers in the field have claimed 1-bit high-sampling format is unsuitable for an audio archival purpose. A quick note is included why this is not the case if Trellis Noise Shaping Converters are used.