Re: LF: WOLF Monkey Business

[Markus Vester <markusvester@aol.com>]

Bill, thanks for the hint! A couple of days ago, I stumbled across this problem but didn't get to the bottom of it. I had attempted to evade a slight difference between the netbook's input and output samplerate at nominal 11.025. I tried 48000, 24000 and 8000 in WolfGui, and each time the output audio was low pitched. I ended up properly calibrating input and output SR separately, entered the results (11025.387 and 11024.770 in my case) into the config, and all was ok.

 

Best 73,

Markus

-----Ursprüngliche Mitteilung-----
Von: Bill de Carle <ve2iq@magma.ca>
An: rsgb_lf_group <rsgb_lf_group@blacksheep.org>
Verschickt: Di, 31 Jan 2012 3:44 am
Betreff: LF: WOLF Monkey Business

I have here two versions of the original WOLF command-line program:
0.51 and 0.61 - apparently there is also a version 0.62 which is used
by the GUI program.

I wanted to duplicate what Stefan did using the GUI version: generate
a signal on 8970 Hz using 24000 s/s.  It *seemed* to work OK but when
I examined the sound card output signal with Cool Edit, the frequency
was way low - about 8241 Hz instead of 8970.  So I went back to the
original command line programs.  I found that versions 0.51 and 0.61
apparently worked, generating 2,304,000 audio samples (exactly 96
seconds at 24000 s/s) to a file called  WOLFX.WAV.  Unfortunately the
samplerate in the .wav file header was coded as 8000 s/s, which means
if we played it back it would run for 288 seconds and the audio
carrier frequency would appear at 2990 Hz.  So I went in and manually
changed the s/s field in the wav header to 24000 s/s.  Then the file
worked OK.  I was able to decode the message by mixing the 8970 Hz
down to 800 Hz and decimating the sample rate to 8000 s/s.  Turning
again to the GUI version, I asked it to encode a .wav file at 24000
s/s with a carrier freq of 8970 Hz.  It reported version 0.62 and
everything OK.  However, when I examined the WOLFX.WAV file so
created, its .wav header said the samplerate was 22050 s/s instead of
24000 s/s!  That explains why the real-time GUI uutput seemed low in
frequency: 8970 * 22050 / 24000 = 8241.1875 Hz.  The upshot is we
need to manually check the wav header and set the s/s field to 24000
instead of whatever else it might say.  And furthermore, don't trust
the real-time Tx soundcard output signal frequency with the GUI
version, especially when using odd-ball sample rates and high output
frequencies which may never have been tested.  Be aware that WOLF
generates envelope-shaped audio, so it might be more difficult to
convert it to a clean square wave to drive a Tx, especially near the
phase transitions where the amplitude drops down to zero.

73,
Bill VE2IQ