A scalable subband audio coding scheme for ATM environments
Wen Jin, Michael S. Scordilis · 2002
A scalable audio coder is proposed for maintaining uninterrupted service in dynamically changing ATM network environments. Performance requirements include seamless switching between bit rates, which would gracefully vary the perceived signal quality. The encoder is similar to the MPEG-1 Layer 3 (MP3) standard but differs significantly in that the MDCT coefficients are quantized in a layered manner. Four enhancement layers that progressively address residual quantization errors follow a perceptually optimized core layer with somewhat poor quantization resolution. Each layer generates its own independently decodable bit stream. Core layer bits appended with enhancement bits are packed into one frame. Signal quality gracefully varies from basic to transparent with the corresponding bit rates of 8, 16, 32, 64, and 128 kb/s at each layer. Subjective tests indicate that our codec performs equivalently to MP3 at 128 kb/s with a somewhat inferior performance at lower bit rates.