Research Of G.729 Speech Coder Implementation Based On DSP

Speech communication is one of the most basic and important way to human-being.The appearance of telephone is the landmark of speech communication it takes us gigantic convenience in our life.In the past half century we have been achieved fast progress on speech codeing technologies.Sientists put forward lots of excellent speech codeing algorithms and did huge contribution.Along with development of information transport,speech coding is playing more and more important roles.Especially the low-bit-rate and high quality codeing of speech,it can provides high quality multimedia service in circumstance of limited bandwidth.And G.729 is one excellent protocol among them.In 1996,the ITU-T standardrized a codeing of speech at 8kbps using S-ACELP(Conjugate-Structrue Algebraic-Code-Exited Linear-Prediction)as Recommendation G.729.For its high quality, G.729 Recommendation is widely used in mobile-communication, satellite communication.multimedia communication in now days and also expected to be use in 3G Personal mobilephone system as standard algorithms which encode speech signal and decode it.It is also worth to mention here that G.729 is hot point on multimedia services such as VoIP and viewphone-meeting.The first general DSP chip was put out in 1980s,with a hard multiplier which is different from common micro-processor.With the great progress and a series of breakpoint in VLSI technologies,in now days as a special element to digital signal processing,DSP is widely used on many field such as multimedia,telecommunication.Because DSP possess many advantages,such as compactness,low power cost,high working speed and low price,it is very suitable for realtime implementation on speech and video compression.Among various DSP products,the TMS320C54X series DSP chips produced by TI company is fitted to the realtime implementation of G.729 speech compression.This thesis particularly introduce the algorithm structure of G.729,including part of encode and decode. It put emphasis on some key technologies ofCS-ACELP algorithm such as Linear-Prediction,Perceptual Weighting,Vector Quantitization,Gain Quantitization,Adptive-codebook Search and Fixed-codebook Search.It also simply described the work theory and inner structure of TMS320C54X series DSP chips.And based on above,it put forward the blueprint of hard system and the software system.Then using the C source code provided by ITU-T, software simulate experiment was done in order to check the correctness of code.At last the author performs the G.729 with TMS320C5402 DSP and optimize it.This paper consists of 5 parts.The first chapter introduces the state-of-the-art of speech compressing technique. And the second part introduces the characteristics of speech signal and some key technique.The third chapter gives the theory and structure of G.729 algorithm.The fourth part of this paper described the blueprint of hardware system , software system,the implementation on DSP and draws the conclusion, including the code optimization and practice experiences. The final sector summarized the whole works and suggested some improvement that should be done in the next.