Hi all,
My, you've been busy here -- lots of very impressive progress! Many
thanks to all of you contributors to the WSJT-related projects!
[Brief aside: My wife and I thoroughly enjoyed our 8-day cruise --
Venice to Athens, with stopovers at ports on the Dalmatian coast: Split,
Korcula, Dubrovnik, Kotor, Butrint, Corfu, and Delphi; then into the
Ionian sea and through the Corinth canal into the Aegean, ending at
Piraeus. Two canceled flights on the way home extended our trip by more
than 24 hours, and 2 of our 3 bags are currently lost -- but otherwise
all is well.]
Here's a start toward responding to some issues raised in the past two
weeks:
1. I've briefly tried the G4WJS "r5629-dirty" version of WSJT-X -- the
one with Bill's suggested changes to the user interface. They look very
good, and I suggest they should be committed to our -devel branch.
2. Our other "Bill", ND0B, has made great progress with implementing
short-sequence ISCAT capability in WSJT. Bill has a bunch of
enthusiastic testers using it on 6 meters with excellent results. I
haven't tried it yet, but after reading the reports from others it seems
that you must be nearing the point of committing the "v9" code (or
something similar) to the SVN repository. Is that right?
A related question: Bill has started a Yahoo Group (possibly to changed
to a Google Group?) to host discussion among the testers of his
experimental version. No doubt this made good sense in early phases of
the effort; there's a downside, however, to moving away from this list
some important communication among programmers working on WSJT-related
code. If others have views on this matter, please share them here.
3. Related to the above ISCAT developments: Bill (ND0B), do you have a
few example *.wav files illustrating the "RRS/RRT" decoding problem? If
so, could you post them somewhere or send them to me? I'd like to look
into the problem.
4. It's hardly surprising that Charlie (G3WDG) and others have found
that "correlation decodes" (in WSJT-X, presently implemented only for
JT4) can produce different confidence levels and (rarely) even different
message results when run against CALL3.TXT files of very different
lengths. After all, the correlation algorithm is effectively answering
these two questions:
A) Which one of the following list of plausible messages best
matches the tone sequence of the received signal?
B) Is the "best" match better than the "second-best" match by a
large enough margin for us to be reasonably certain we have a valid decode?
Obviously, the answers to both questions will depend on the length of
the list of plausible messages, which is generated from call+grid
combinations derived from CALL3.TXT, augmented by the "DX Call" and "DX
Grid" entries on the main window. If the list is short (but still
contains the call and grid actually in the message), the chances of a
correct decode and the estimated confidence in its validity will be
higher than with a long list.
5. I'm delighted to hear that Steve (K9AN) has implemented a WSPR signal
subtraction algorithm that works so well! I haven't looked at the code
or tried it yet, but from Steve's report it sounds like we should make
wsprd_exp (renamed to wsprd) the default WSPR decoder. Perhaps we can
use the existing "Fast / Normal / Deep" selection on the "Decode" menu
to control whether subtraction and multi-pass decoding will be used, or
not? I'm not too worried about the longer decoding times: as we have
found previously, significant optimizations will likely be possible
after we have it working well. Furthermore, fast decoding is arguably
of minimal importance in WSPR mode, since no quick operator interactions
are required.
6. Signal dropouts in the *.c2 files are a concern -- we'd better find
out what's causing them. An important question to be answered: are the
dropouts present in the c2 array in memory, or just in the file as
written to disk?
That's probably enough for my first day back on the job...
-- 73, Joe, K1JT
------------------------------------------------------------------------------
Monitor 25 network devices or servers for free with OpManager!
OpManager is web-based network management software that monitors
network devices and physical & virtual servers, alerts via email & sms
for fault. Monitor 25 devices for free with no restriction. Download now
http://ad.doubleclick.net/ddm/clk/292181274;119417398;o
_______________________________________________
wsjt-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
