Noise shaping in an ITU-T G.711-interoperable embedded codec

In the transition from narrowband to wideband speech communications, there is a need in some applications for a high quality wideband coding scheme interoperable with the ITU-T G.711 narrowband coding standard. This can be accomplished using a multi-layer coding scheme with a G.711 compatible core layer. For optimal wideband quality in the upper layers, this requires using full frequency range (50-4000 Hz instead of 300-3400Hz) in the core layer. In this context, the 8-bit non-uniform PCM quantizer of the ITU-T G.711 standard can produce highly perceptible noise. The purpose of this paper is to demonstrate how efficient noise masking can be applied at the encoder in a G.711-interoperable manner, and how the same noise masking can be extended at the decoder to one or more enhancement layers to implement a perceptually optimized multilayer codec.