Optimization and Implementation of Speech Codec Based on Symmetric Multi-processor Platform

This paper proposes an efficient implementation scheme of the speech codec based on the embedded symmetric multiprocessor data signal processing platform. The base codec uses the G.729 algorithm. To satisfy the requirements of real time applications for efficient processing both audio and video simultaneously on multi-channels, this paper present some implementation techniques to optimize the speech coder. To exploit the characteristics of the embedded symmetric multiprocessor platform, we propose a coarse-grained functional partitioning method to balance the load of the encoder among the cores with small overhead of synchronization. In order to reduce the computational burdens, we propose a fast voice activity detection method. The experimental results show that the proposed scheme improves the performance of the speech codec greatly in terms of computational complexity.