Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!uunet!mcsun!corton!chorus!opera!mir
From: mir@opera.chorus.fr (Adam Mirowski)
Newsgroups: comp.sys.ibm.pc.hardware
Subject: Re: Analog-to-Digital Sound-to-Data pc board?
Message-ID: <7764@chorus.fr>
Date: 7 Feb 91 11:35:21 GMT
References: <15917@milton.u.washington.edu> <1991Feb6.145754.11445@d.cs.okstate.edu>
Sender: mir@chorus.fr
Reply-To: mir@opera.chorus.fr (Adam Mirowski)
Organization: Chorus systemes, Saint Quentin en Yvelines, France
Lines: 46

In article <1991Feb6.145754.11445@d.cs.okstate.edu>, ong@d.cs.okstate.edu (ONG ENG TENG) writes:

%% The Sound Blaster has a digital sampling 8-bit analog input.  To record
%% voice, you only have to buy a regular microphone with -70 dB or better
%% (i.e. -69, -68,...).  It samples from 4kHz to 12kHz.  If you want lower
%% frequency, simply ignore those in between.  Example to get 1kHz sampling
%% rate data simply sample at 4kHz and use only 1 value out of each 4.
%% The card comes with both menu-driven and line-command programs to
%% record voice.  The voice file generated has a 32-byte header followed
%% by the data in raw format. 

There are two things I think are wrong here:

First: before undersampling a sampled signal you have to put it
through a (digital) filter, to cut higher frequencies. Or you will
get the same effect as sampling an analog signal without respecting
the Shannon rule (sampling frequency must be twice as high as the
highest frequency component of the input signal). Of course, in the
example, if the original signal had an under-500Hz spectrum, it would
be not necessary.

Second: To make a correct .VOC file (SB standard sample file), you
really need to put a null byte at the end. Without this null byte,
the playing utility won't end properly.

Also, if John Siemion needs frequencies like 12543, 12501, etc.
SB will definitely not work, because the sampling frequency
should be at least 24-25 KHz then. Secondary question: why
isn't SB able to sample at a higher frequency? :-)

Another thought: does John need to SAMPLE a signal or to MEASURE
its instantaneous frequncy? Because in the latter case, you
should do something like series of FFT on the sample before
getting the frequency. That is a lot of computation.

Maybe a simple circuitry build around a PLL loop at 12KHz could
provide an error voltage "proportional" to the difference in
frequency between the input signal and 12KHz.  You could then
sample this voltage with an SB or a simpler board.  There are
several one-chip PLL implementations. Of course, all that is
only good if your input signal is really around some fixed
frequency.

-- 
Adam Mirowski,  mir@chorus.fr (FRANCE),  tel. +33 (1) 30-64-82-00 or 74
Chorus systemes, 6, av.Gustave Eiffel, 78182 Saint-Quentin-en-Yvelines CEDEX