Hi Bill, Joe, and all - 

For what it’s worth — while working on my decoder, I found that this test (in 
routine extract.F90) is rejecting a lot of decodable symbol vectors:

  if(ntest.lt.100) then
     ncount=-999                      !Flag and reject bad data
     go to 900
  endif

If I disable this test, the yield increases by almost a factor of two. I also 
find that I can now decode many of the “missed” decodes that are listed below 
(using either sfrsd or kvasd).

Note that in the older python code contained in the trunk, the ntest threshold 
is set to 50, so this could explain some of the difference between the current 
implementation of jt65 and older ones.

I stumbled on this because it turns out that my decoder is going to need to use 
different soft symbol metrics than kvasd does. When I tried to use my metrisz, 
every single vector was rejected by the ntest check — so I needed to remove it 
for my testing purposes. That’s when I noticed that it significantly increased 
the yield from kvasd as well.

Steve


> On Sep 16, 2015, at 11:17 AM, Bill Somerville <g4...@classdesign.com> wrote:
> 
> Hi All,
> 
> I am discussing with a couple of OMs issues they have with JT65-HF 
> consistently decoding better on weak signals. The common factor is that 
> the KV decoder is used on JT65-HF (it marks that with a K on the decode 
> like WSJT does).
> 
> Here are a selection of .WAV files:
> 
> https://dl.dropboxusercontent.com/u/4192709/WSJT-X%20decoder%20fails%20samples.zip
> 
> the details of the missed decodes are:
> 
> 14:30 15 K -22 -0.1 65 DJ6JZ XE2PXN -10 XE
> 14:30 15 K -21 1.1 -188 PA0LUX VU2DPN -14 VU
> 
> 14:33 15 K -24 -0.1 -834 SP1TMH ZS1AW 73 1 Band
> 
> 15:29 15 K -23 1.3 -323 CQ KV4HV EL94 > EL94
> 
> 15:30 15 K -22 0.6 -770 DJ6JZ N4LPD -01
> 
> 15:31 15 K -22 1.3 272 CQ KV4HV EL94 > EL94
> 15:31 15 K -17 0.8 -250 OZ5KAJ PU1OHM GG87 > GG87
> 
> The DF figures are relative to 21077.5.
> 
> Any ideas? Is the recently discussed birdie removal algorithm prior to 
> calling KVASD causing this?
> 
> 73
> Bill
> G4WJS.
> 
> ------------------------------------------------------------------------------
> Monitor Your Dynamic Infrastructure at Any Scale With Datadog!
> Get real-time metrics from all of your servers, apps and tools
> in one place.
> SourceForge users - Click here to start your Free Trial of Datadog now!
> http://pubads.g.doubleclick.net/gampad/clk?id=241902991&iu=/4140
> _______________________________________________
> wsjt-devel mailing list
> wsjt-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/wsjt-devel


------------------------------------------------------------------------------
Monitor Your Dynamic Infrastructure at Any Scale With Datadog!
Get real-time metrics from all of your servers, apps and tools
in one place.
SourceForge users - Click here to start your Free Trial of Datadog now!
http://pubads.g.doubleclick.net/gampad/clk?id=241902991&iu=/4140
_______________________________________________
wsjt-devel mailing list
wsjt-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/wsjt-devel

Reply via email to