Perceptually Layered Scalable Codec

Fine-grain scalable audio coding approaches, such as PLEAC, are flexible, can scale from low rates to lossless, and can allow the perceptual model to be updated in the decoder as additional bits are provided. They provide good perceptual quality at mid to high perceptual rates; however, they can not establish the perceptual model rapidly enough to provide good quality at very low bit rates. In this work, we combine a fine grain scalable audio codec with a good base layer audio codec. We use Windows Media Audio 10 Professional (WMA 10 Pro) as the base layer coder, and use the PLEAC codec as the fine grain enhancement codec. The key to this method is to use a base layer codec that provides a very good approximation to the necessary perceptual codec, and then allows the fine grain scalable audio coder to use a simple loudness model on the difference signal. The combined codec can provide continuous scalability from the base layer codec rate all the way to lossless.