Hi Steinar,
you wrote:
Wolf,
Did you get my audio file ?
LA5VNA S
Yes, many thanks for that.. maybe a previous message didn't get out, or
was lost on the way...
>>
Ok Steinar,
indeed, completely downloaded now. Maybe I was too quick when trying,
and the file wasn't complete on the server yet.
I already tried Rik's large wave file, and getting a good decode from
his reception of Domenico. Maybe it's the same strong carrier around
475.2 kHz (which made copying with the original decoder impossible here).
Since there is no 'intelligent' automatic notch integrated yet, it's
important to adjust the decoding frequency range so that there is no
"carrier" visible between the red (or yellow, for the 2nd channel)
markers in the spectrogram.
With that, Rik's wave file produced the following:
cq cr bq eeiz7tk iz7slz iz7slz
Ok, far from being perfect, but -similar to my favourite mode- it's not
necessary to copy each character 100% error-free. That's part of the
fun, as well as tweaking the parameters.
Decoding Steinar's 33 MByte file, with SNR between -15 and -25 dB
(towards the end):
--- Replay
'C:\WaveArchive\WSQ_Tests_2014\2014_03_05_LA5VNA_WSQ2_no_decode.wav' ---
c iy8slz de dl4yhf
-12 to -25 d\
rrr fb but here some errors now e
own padh ng 3 NBysf frol Sij.@
tryhog l uds a
n xudtt a d bvk,@h.f@tlzd l4yhf|3 s
There seemed to be some spectral spreading at the end, irritating the
simple peak-searching loop in the decoder.
But with this file (along with Rik's and my own recordings of IZ7SLZ) I
can now try to modify the decoder and see if a different method for
finding the peak frequencies (equivalent to symbols in WSQ) can improve
the behaviour when multi-path 'widens' the signal.
All the best,
Wolf .
|
