| Summary: | 碩士 === 淡江大學 === 電機工程學系 === 84 === The speech application are very popular in the world today,
such as telephone,voice mail box,speech recognition,speech
synthesis...etc.The amout of orginal speech data is very large.
If we don't process those stored speech data , it must cost much
memory.For example,if we record speech signal with 8 KHZ
sampling rate and 8 bits per sample,we need 64k bits memory
per second and 3.84 Mbits per minute . In the region of speech
coding, people try to reduce the bit rate of speech data and
make reconstructed signal comfortable for people.This paper
focuses on using less bit rate to implement speech application.
Sound baster is used to record speech signal and Adaptive
Differential Pulse Code Modulation (ADPCM ) bases on
International Telegraph and Telephone Consultative Committee
(CCITT) G.721 standard is adopted to compress speech data in
this paper.The sampling rate is 8 KHZ and 4 bits per sample. In
ADPCM system , one word memory adaptive quantizer and adaptive
preditor is adopted to reduce bit rate in this paper. In the
decoder of ADPCM , a 8-bit mircocontroller is adopted to
implement this algorithm.Besides,the paper use verilog to
simulate the interface between the mircocontroller,memory and D-
to-A converter.After finishing all simulation,the paper uses
FPGA to verify whole system in real world. Besides,this paper
design a simple application using this speech decoder to make
sure the quantity of output speech accepted by the people.
|
|---|
