The Sound Signal Feature Parameters' Extraction Of Covariance Lattice Algorithm On FPGA

With the development of speech technology application, more and more speech-signal digital processing system has to be real-time mode or to be on-line mode. The requirement of hardware environment is higher and higher, especially in the design of embedded system. The speech-processing arithmetic is more and more complicated, and ability not equal to our ambition that to deal with speech-signal real time with common processor. Special speech-chip can resolve the problem of parallel and speed of speech-signal processing, so the application on the speech-signal processing system is especially important.Using the predominance of new Altera programmable logic chip on digital signal processing domain, lucubrate on the implement with FPGA(Field Programmable Gate Arry) of the extraction of speech-signal's common parameter-LPC parameter. Firstly, analyze the speech disperse digital model and short-time smooth characteristic, discuss speech Linear Prediction Coding (LPC) in-depth. Secondly, compared the solution of linear prediction system of equations of self-correlation algorithm and covariance lattice algorithm, and bring forward a collectivity scheme of LPC character parameter extraction system based on covariance lattice algorithm. Thirdly, study the in-side structure of Altera Cyclon series programmable chip, and study the flow of FPGA design on the Quartus II exploitation flat roof. Fourthly, design and implement two function module, self-correlation and Durbin arithmetic, analyses and explain the work process in detail. Fifthly, the characteristic of division operation in this system is that quotient is less than 1, so design and implement a special division module. Finally, on the aim chip Altera FPGA EP1C6Q240C8, emulate and validate the LPC character parameter extraction system. System has the characteristic of fast processing, besides, has flexible input and output interface. This system can be joined with other speed-processing module expediently, composing an entire speech-processing system on a chip, and can be used in speech coding and decoding system, speech synthesize system, speech recognizing system and so on.