Content-Transfer-Encoding: 8bit
Message-ID: <3617AC3245C2D1118A840000F805359C01AB8CD9@pdw-mercury-1.dera.gov.uk>
From: "Talbot Andrew" <ACTALBOT@dera.gov.uk>
To: rsgb_lf_group@blacksheep.org
Subject: LF: RE: Shrinking sounds - fiasco
Date: Fri, 23 Jun 2000 10:27:59 +0100
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1; format=flowed
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>


You will not be able to shrink files and make them audible without
processing the .WAV file data considerably.  Merely changing the
sampling rate will do nothing to improve readability.   The stages
necessary are  :

1)	Take the off air signal recording at (say) 8000 Hz sampling rate

2)	Digitally mix down to some arbitrarily low frequency by
multiplying each waveform sample by COS and SIN samples of a digital
local oscillator signal.

3)	Filter to the necessary sub-Hz bandwidth centred on the signal
of interest .  This stage is what makes the signal 'audible' and brings
it out of the noise.

4)	Decimate this filtered data to a much lower sampling rate by
just taking one sample out of every (say) 256 . Save as if it were a
.WAV file giving the data sampled at 31.25 Hz.  Choose a decimated
sampling rate adequate to meet the Nyquist criteria of the bandpass
filtered signal

5)	Re label the .WAV file to make it 'think' the sampling rate is
8000Hz and replay.  The result will be a tone centred on 256 * the mixed
down frequency with a bandwidth 256 * the filtered bandwidth.

All the numbers can of course be changed arbitrarily provided teh rules
for sampling and Nyquist are adhered to.  As you can see, the process is
not as straighforward as some have lead to believe on this reflector.  
For anyone contemplating writing software to do this, note that stages
3) and 4) can be taken together - if using an FIR filter, the filtering
process needs to be done for each decimated sample only - this saves
considerable processing overhead for the long filter tap lengths
necessary

I have software to do this process for other (non-amateur) signal
analysis purposes and use an intermediate I/Q file format where signals
are first mixed down to zero centre frequency and stored in their
complex form with a spectrum ranging from minus to plus half the
sampling rate.  Processing is then more straightforward as bandpass
filtering becomes lowpass etc.  After decimation a final conversion to
.WAV format is then made and the tone frequency and false sampling rate
can be set arbitrarily.

The most fascinating signal produced this way was a Dopplergram of a
carrier on 3.572MHz recorded overnight and played back 500 times faster.
The ionospheric Doppler shifts and multipath result in sounds not unlike
those of Whales talking !

Andy  G4JNT
> ----------
> From: 	Petr Maly[SMTP:p.maly@gmc.net]
> Reply To: 	rsgb_lf_group@blacksheep.org
> Sent: 	2000-06-23 08:56
> To: 	136 group
> Subject: 	LF: Shrinking sounds - fiasco
> 
> Hello all
> Yesterday I spent a while with shrinking WAV's in SoundForge and
> CoolEdit.
> In the original file (da0lf), there is nothing heard by ear, but when
> the
> file is open in the Gram, the signal trace is clearly visible. I tried
> to
> shrink to 50% of the previous length as a first step. I could not hear
> anything either. When I opened the new file in the Gram, the signal
> disappeared completely. I guess the main reason is that unlike the
> Bill's
> method, this one doesn't lead to narrowing the bandwidth. Simply, it
> doesn't
> work. I must try Bill's software.
> 
> 73, Petr, OK1FIG
> 
> 
> 
> 


-- 
The Information contained in this E-Mail and any subsequent correspondence
is private and is intended solely for the intended recipient(s).
For those other than the recipient any disclosure, copying, distribution, 
or any action taken or omitted to be taken in reliance on such information is
prohibited and may be unlawful.