Research And Realization Of G.729 Speech Codec Algorithm Based On DSP

Along with the development of information transport in recent years, speech coding is playing more and more important roles. Especially the low-bit-rate and high quality coding of speech, it can provides high quality multimedia service in circumstance of limited bandwidth. In 1996, ITU-T issued the G.729 Recommendation. It is used for coding at 8kb/s speed by Conjugate-Structure Algebraic-Coder-Excited Linear-Prediction (CS-ACELP) algorithm, which has a 15ms delay. It has nicer synthesized speech quality and appropriate complexity. It can be widely used in the field of individual mobile communication, multimedia communication, IP phone, satellite communication, ISDN and so on.With the development of digital signal processing technology, the application of DSP is more and more extensive. Because it possess many advantages, such as compactness, low power cost, high working speed and low price, it is very suitable for realtime implementation on speech. The TMS320C55x series DSP chips produced by TI company is fitted to the real time implementation of G.729 speech compression.This paper particularly introduce the algorithm structure of G.729, including part of encode and decode. It put emphasis on some key technologies of G.729 algorithm such as Linear-Prediction, Perceptual Weighting, Vector Quantization, Adptive-codebook Search and Fixed-codebook Search. It also simply described the work theory and inner structure of TMS320C55x series DSP chips. Based on grasping the principle of G.729, design techniques are presented for efficient implementation of G.729 standard. Then using the C source code provided by ITU-T, software simulate experiment was done. It optimizes the larger operational quantity modules of the source code based on the characteristics of G.729 and finally achieves the real time implementation of speech codec based on DSP. The result shows the synthesized speech has better intelligibility.