Hi Joe, My post is not a complaint, simply a beta report on the software. The number of false decodes I'm seeing is much fewer than 1%, and their printed content corresponds to your description. I agree that's quite acceptable if it also provides additional good decodes. It might, however, be a good idea to add a sentence to the user doc about that if it isn't already there. I just printed it last night, but haven't collated or read it yet.
If you can tell me where to find the .wav, I can send it, but I'm not sure if you would want it. I can't even find the file corresponding to the left display of decodes. :) Thanks and 73, Jim K9YC On Thu,12/15/2016 7:12 AM, Joe Taylor wrote: > Hi Jim, > > Without specifics and saved .wav files it's impossible to comment in any > detail on your false decodes. > > When developing the two-pass decoder, we analyzed over a thousand .wav > files in crowded band conditions. The probability of false decodes was > higher for second-pass decodes than for first-pass decodes, but was > still less than 1 percent. We considered that an acceptable price for > the big advantage of many good decodes of very weak, otherwise buried > signals. > > False decodes are almost always easy to recognize as such. Roughly half > should be random, garbled free text. The other half will contain things > that look vaguely like two callsigns [or CQ/QRZ/DE ... plus one > callsign], and probably a locator. Usually one or both of the > "callsigns" will have a nonstandard, bogus-looking prefix. There will > *not* be any correlation between the apparent prefix and the locator. > > -- 73, Joe, K1JT > > On 12/15/2016 3:04 AM, Jim Brown wrote: >> Hi Team, >> >> For several weeks, I've been letting WSJT-X rc3 run overnight on 160M >> while I sleep, and tabulating the decodes the next day. I've been quite >> pleasantly surprised by the activity -- in less than a month, I've >> decoded about 450 different stations running JT65 or JT9. More than a >> dozen are from outside North America, about half of those from western EU. >> >> The reason for this post is that I've also seen a half dozen or so >> decodes that appear to be false. Looking more closely, I see that most >> are second decodes finding the signal underneath another one that >> decodes correctly. >> >> I haven't done enough analysis to say that ALL second decodes are bogus, >> but nearly all of the bogus decodes are on second passes. On each of >> them, I can't find the call on qrz.com, the call often doesn't make >> sense, and the time of day doesn't support propagation to me (near San >> Francisco) from the received grid square. BUT -- the grid square usually >> roughly correlates with the country associated with the call (which is >> usually a non-standard prefix for that country). >> >> 73, Jim K9YC >> >> >> ------------------------------------------------------------------------------ >> Check out the vibrant tech community on one of the world's most >> engaging tech sites, SlashDot.org! http://sdm.link/slashdot >> _______________________________________________ >> wsjt-devel mailing list >> wsjt-devel@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/wsjt-devel >> > ------------------------------------------------------------------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, SlashDot.org! http://sdm.link/slashdot > _______________________________________________ > wsjt-devel mailing list > wsjt-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/wsjt-devel > ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot _______________________________________________ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
