A Hardware Accelerated MP3 Decoder with Bluetooth Streaming Capabilities

Digital music players have been on the market for several years. Most of these players store compressed music in non-volatile memory such as memory cards or flash. The players have all been constructed focusing on MP3 decoding, using specialized hardware instead of a general design. Nowadays, handheld computers have enough computational power to decode compressed music in real-time. This solution is however not power efficient. Proposed in this paper is a new general decoding concept combining high performance with low power consumption. Also presented in this paper is a solution using Bluetooth data streaming instead of non-volatile memory. Combining Bluetooth streaming with a general decoding hardware, a highly optimized MP3 decoding engine has been implemented, complying with the high ISO/IEC 11172-3 quality demands. This is achieved by designing a hardware accelerator, focused on calculating Discrete Cosine Transforms (DCT) and convolutional algorithms. This general design also has the advantage of being able to accelerate and support additional compression algorithms, such as other music or even image formats. The computation speed of the most critical decoding steps are improved with 300% using only 18% of the power consumption and adding only 10% to the total area compared to using a small microprocessor based solution. Acknowledgement The authors would like to thank the employees at the ASIC-department of C-Technologies AB, especially Tomas Edsö and Niclas Bengtsson for their continuous support on both the work and the documentation. Also thanks to Anders Berkeman and Viktor Öwall at the department of Electroscience, Lund University, for valuable help. Special thanks to Johan Fredriksson for helping out with the CAD work.