Voice Compression And Expansion Algorithms And Hardware Realization

The purpose of researches of speech technology are all for applications. Forexample, the research of transition and store of digital speech mostly enhanced thereliability, anti-jamming, high-speed switch and secrecy of speech transition and storesystem. Otherwise, it reduced the cost of the systems. Speech coding provided thepossibility for realization of narrowband communication system and drove theapplication of low-cost speech store system and speech synthesis system.In this paper, we study the theory of speech coding, and give a new, simple andrealizable speech coding algorithm for our application.The purpose of this paper is designing 4 second length speech synthesis ASIC.This ASIC is mostly used in some speech products, so there are many characteristicsin the speech coding algorithm we have to design: no transition problem, no real-timedemand, original input speech can be optimized for our need, the demand ofsynthetical speech quality is high and the low-cost of speech synthesis ASIC.In this paper, we give a new speech coding algorithm based on ADPCM. Thisalgorithm is just for the design of 4 second speech synthesis ASIC. The importantcharacteristics of this new speech coding algorithm are: integer data type, shiftoperation for division, high performance of quantity and simple predictor.By the simulation of C language procedure, we conclude that the quality ofsynthesis speech is very good, because it is mostly same with the original input speech.And that the area, the power efficiency and the quality of synthetical speech of 4second length speech synthesis ASIC all meet the demand of actual application.