We describe a perceptual audio coder that employs nonuniform hierarchical filterbanks and entropy-coded lattice vector quantization. The nonuniform filterbanks present a good match to critical-band sampling, and they avoid the complications of block-size adjustment during signal transients. An entropy-coded lattice vector quantizer provides efficient quantization and coding of the filterbank samples. The coder achieves “transparent” quality at a rate of 96 kb/s, and efficient algorithms exist for the filterbank and vector quantizer subsystems
Published in:
Acoustics, Speech, and Signal Processing, 1994. ICASSP-94., 1994 IEEE International Conference on
(Volume:ii
)
Date of Conference: 19-22 Apr 1994