MPEG-2 audio decoding algorithms are involved of several complex coding tec
hniques and therefore difficult to be implemented by an efficient dedicated
architecture design. In this paper, we present an effective architecture f
or the MPEG-2 audio decoder. The MPEG-2 audio algorithms can be roughly div
ided into two types of operations. The first type is regular but computatio
n-intensive such as filtering, matrixing, degrouping, and transformation op
erations. The second type is not regular but computation-intensive such as
parsing bitstream, setting operation mode and controlling of all system ope
rations. A RISC core with variable instruction length is designed for the d
ecision-making part, and the dedicated hardware such as special divider, an
d synthesis filterbank is designed for the computation-intensive part. Base
d on the standard cell design technique, the VLSI chip consists of 27000 ga
te counts with the chip size of 6.4 x 6.4 mm(2). The chip can run at maximu
m 43.5 MHz clock rate, with the power dissipation of 150 mW at 3 V power su
pply.