Audio playback requires long battery life in handheld applications like portable audio players. To achieve low power and die area, the number of gates needs to be reduced. An audio coprocessor can help in achieving low power and die area. The audio coprocessor/accelerator needs to be programmable to address multiple codec standards (new and existing), and numerous pre- and post-processing algorithms. This paper describes assembly programming techniques of an audio coprocessor for the MP3 decoder for better utilization of hardware and the optimization of cycles and memory.