🚀 go-pugleaf

In article <1gg7rlINNcs1@charnel.ecst.csuchico.edu>, helho@ecst.csuchico.edu (Weird!!) writes:
|> Hi,
|>    I am trying to build a project which is able to digitise human
|>    voice by using a microphone input and a small speaker output.
|>    Since the band width of human voice is no more than 10KHz ( i guess)
|>    ,therefore I need a antialiasing filter to cut off those unused   
|>    frequency ( is that right?) What kind of filter is good enough to
|>    deal with this job? Then the filtered analog voice signal will be 
|>    digitized by a A/D converter with a sample-hold built-in. My choice
|>    is National semiconductor DA8020, is it a good choice? 
|>    Then The 8 bits digital signal will go to a D/A converter( DC8010 )
|>    Is it a good choice? The analog output will connect with a smoothing
|>    filter to "smooth" the signal. Finally the message will be played
|>    back by a 8ohm/4ohm speaker.
|>    Now, my questions are :
|>      (1) Are those A/D & D/A  good enough for voice analysis?
|>      (2) Which antialiasing and smoothing filter is suitable?
|>      (3) How to interface the microphone with the circuit?
|>      (4) How to connect the speaker with D/A converter?
|>      (5) sampling rate = input clock signal ?
|> 	 

Those chips should be adequate for your needs.  You probably only
need a bandwidth of 5KHz or so.  You should build a 5KHz low pass filter
on the input, and then sample at 10KHz.  In fact, you could probably
do a 4KHz low pass filter and sample at 8KHz without losing much
information.  Just remember to filter at 0.5 of your sample rate.

You probably will not neet to smooth the output very much.  The 
coil in the speaker will take care of most of it.  You may want to
add a capacitor across the speaker.

-- 
Larry D. Pyeatt                 The views expressed here are not
Internet : pyeatt@texaco.com    those of my employer or of anyone
Voice    : (713) 975-4056       that I know of with the possible
                                exception of myself.