Call Search
     

New to Ham Radio?
My Profile

Community
Articles
Forums
News
Reviews
Friends Remembered
Strays
Survey Question

Operating
Contesting
DX Cluster Spots
Propagation

Resources
Calendar
Classifieds
Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement
About eHam.net

   Home   Help Search  
Pages: Prev 1 [2] 3 Next   Go Down
  Print  
Author Topic: New Bayesian Morse decoder available for alpha testing  (Read 11952 times)
AG1LE
Member

Posts: 119


WWW

Ignore
« Reply #15 on: January 03, 2014, 07:02:02 PM »

I can't test this anymore. I screwed something up when I was playing with the mixer.

Maybe someone can give me some advice. I don't know where to begin to fix this.

I was trying to find the setting in the mixer that would let me adjust the signal level to running applications. In the process, I managed to "kill" the audio the signal to any running applications, including fldigi and qsstv.

The sound from my rig still comes out the computer speakers, as it always has. But that's all.

I restored the old /home/mike/.config/xfce4/xfconf/xfce-perchannel-xml/xfce4-mixer.xml but it didn't fix it, even after a reboot. (/home/ is the only backup I have.)

Any suggestions?  I'm using Xubuntu 12.04.3 LTS (Precise), which is Ubuntu with the XFCE 4.8 GUI.


I don't know exactly what Bayesian means, either. I'll concern myself with that after this audio issue is solved. :-)

Have you tried using "pavucontrol" ?   I switched from Ubuntu to Linux Mint  15 for my development laptop and pavucontrol seems to work in more reliable manner than other audio mixers. You can install it  with command   "sudo apt-get install pavucontrol".


73
Mauri AG1LE 
Logged
W0BTU
Member

Posts: 1579


WWW

Ignore
« Reply #16 on: January 04, 2014, 10:14:58 PM »

Have you tried using "pavucontrol" ?   I switched from Ubuntu to Linux Mint  15 for my development laptop and pavucontrol seems to work in more reliable manner than other audio mixers. You can install it  with command   "sudo apt-get install pavucontrol".

Thank you. I installed this, but nothing I do there seems to restore the input to sound recorder, which I had running to test this.

I posted the following to http://ubuntuforums.org/showthread.php?t=2197696&p=12891761#post12891761
Stumped! Sound from speakers, but can't record.

    I was playing with the XFCE Mixer, trying to add/remove controls, and all of a sudden I have no input to Sound Recorder (and other application that require sound input, such as Fldigi and Qsstv). I have no idea what happened, and I am at a loss to fix this.

    I was trying to find the setting in the mixer that would let me adjust the signal level to running applications. In the process, I managed to "kill" the audio the signal to any running applications, including fldigi and qsstv.

    The sound from my rig still comes out the computer speakers, as it always has. But that's all. There's no longer any input to any running application that takes sound input.

    I ran alsamixer and turned everything on.

    I've looked through many threads here and Googled this problem. No luck. I am stuck.

    I restored the old /home/mike/.config/xfce4/xfconf/xfce-perchannel-xml/xfce4-mixer.xml but it didn't fix it, even after a reboot. (/home/ is the only backup I have.)

    Any suggestions? I'm using Xubuntu 12.04.3 LTS (Precise), which is Ubuntu with the XFCE 4.8 GUI.

    Any help would be greatly appreciated! Let me know what else I should post

Logged

AG1LE
Member

Posts: 119


WWW

Ignore
« Reply #17 on: January 25, 2014, 12:17:40 PM »

I posted a new release of new Bayesian Morse Decoder on Dave's web site
http://www.w1hkj.com/ag1le/

File name is fldigi-3.21.77.bayes.tar.gz 

The "pmax" bug is now fixed in this version.

I converted the software from C  to C++ to enable multiple decoder instances in the future.
I also expanded the automatic speed adaptation range  from 10...60 WPM  to  8...80 WPM,
and included Squelch function to disable the decoder when signal level is below set value.

I tested this version with WAV files recorded from ARRL Straight Key Night - I am getting decent decoding accuracy on hand keyed Morse signals. 

Let me know how this new version works for you.

73
Mauri AG1LE 
Logged
W0BTU
Member

Posts: 1579


WWW

Ignore
« Reply #18 on: January 25, 2014, 08:39:33 PM »

I posted a new release of new Bayesian Morse Decoder ... Let me know how this new version works for you.

This sounds interesting, and I would love to test it.

I am afraid to install it, though, because the last time I installed your version it overwrote the other version.

I need to figure out how to install more than one version of fldigi independently. Any suggestions?
Logged

AG1LE
Member

Posts: 119


WWW

Ignore
« Reply #19 on: January 26, 2014, 05:35:23 AM »


This sounds interesting, and I would love to test it.

I am afraid to install it, though, because the last time I installed your version it overwrote the other version.

I need to figure out how to install more than one version of fldigi independently. Any suggestions?
You can download the tar  file to another directory.
Then  execute the following commands (in Linux shell):   

tar  xvf  fldigi-3.21.77.bayes.tar.gz
cd fldigi-3.21.77.bayes
./configure
make

After compilation you should have executable version in src sub-directory.
You can start  it with command "src/fldigi"  while in fldigi-3.21.77.bayes directory.

If you do the normal  "sudo make install"  it will overwrite your existing version.
This way you can have multiple versions that don't interfere each others.

I had this new version running 3 hours last night while I was listening various CW contest stations.
Seems to be much more stable now...I did not get any "pmax" errors and decoding accuracy was quite good.

73
Mauri AG1LE



Logged
AG1LE
Member

Posts: 119


WWW

Ignore
« Reply #20 on: January 29, 2014, 04:01:22 AM »


This sounds interesting, and I would love to test it.

I am afraid to install it, though, because the last time I installed your version it overwrote the other version.

I need to figure out how to install more than one version of fldigi independently. Any suggestions?

Hi Mike

Any luck installing this new version?  Have you had time to give it a try yet?

73
Mauri AG1LE
Logged
G0UKB
Member

Posts: 1




Ignore
« Reply #21 on: February 24, 2014, 04:32:34 AM »

Hi Mauri - just to let you know I've built your version (on Mint 16) and have successfully decoded. On first impressions it certainly seems better than the previous built-in decoder for FLDIGI. in a couple of weeks there is the UK Commonwealth contest (BERU) which is CW and restricted to UK and our commonwealth countries so it is a fairly relaxed contest with no huge pile-ups since Europe and the USA can't enter. I hope to enter this year with CW treated purely as any other data mode on FLDIGI using your decoder. Last year I cobbled together a solution using WD6CBF's CW Decoder on Windows with a logging program since WD6CNF's decoder seemed the best in tests.

Having a decent decoding engine inside FLDIGI will, of course, make this so much easier this year. Should be interesting.

Thanks for the huge amount of work that must have gone into this project and should you find any bugs or improvements then I'd love be on the test list.

Brian G0UKB
Logged
N3HEE
Member

Posts: 113




Ignore
« Reply #22 on: February 24, 2014, 04:10:48 PM »

Errors trying to compile and test.  Followed your instructions above and get the following output from make....  Please help !  Thanks -Joe n3hee

 make
Making all in po
make[1]: Entering directory `/home/joe/fldigibayes/fldigi-3.21.77/po'
make[1]: Leaving directory `/home/joe/fldigibayes/fldigi-3.21.77/po'
Making all in doc
make[1]: Entering directory `/home/joe/fldigibayes/fldigi-3.21.77/doc'
make[1]: Nothing to be done for `all'.
make[1]: Leaving directory `/home/joe/fldigibayes/fldigi-3.21.77/doc'
Making all in src
make[1]: Entering directory `/home/joe/fldigibayes/fldigi-3.21.77/src'
make  all-recursive
make[2]: Entering directory `/home/joe/fldigibayes/fldigi-3.21.77/src'
make[3]: Entering directory `/home/joe/fldigibayes/fldigi-3.21.77/src'
  CXX    fldigi-xmlrpc.o
In file included from ./include/Fl_Text_Editor_mod.H:37:0,
                 from ./include/FTextView.h:34,
                 from ./include/FTextRXTX.h:28,
                 from ./include/fl_digi.h:34,
                 from misc/xmlrpc.cxx:58:
./include/Fl_Text_Display_mod.H:111:5: error: ‘Fl_Fontsize’ does not name a type
./include/Fl_Text_Display_mod.H:262:3: error: ‘Fl_Fontsize’ does not name a type
./include/Fl_Text_Display_mod.H:268:17: error: ‘Fl_Fontsize’ has not been declared
./include/Fl_Text_Display_mod.H:477:3: error: ‘Fl_Fontsize’ does not name a type
./include/Fl_Text_Display_mod.H: In member function ‘void Fl_Text_Display_mod::textsize(int)’:
./include/Fl_Text_Display_mod.H:268:33: error: ‘textsize_’ was not declared in this scope
make[3]: *** [fldigi-xmlrpc.o] Error 1
make[3]: Leaving directory `/home/joe/fldigibayes/fldigi-3.21.77/src'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory `/home/joe/fldigibayes/fldigi-3.21.77/src'
make[1]: *** [all] Error 2
make[1]: Leaving directory `/home/joe/fldigibayes/fldigi-3.21.77/src'
make: *** [all-recursive] Error 1
joe@ubuntu:~/fldigibayes/fldigi-3.21.77$
Logged
AG1LE
Member

Posts: 119


WWW

Ignore
« Reply #23 on: February 24, 2014, 06:41:31 PM »

I googled the error and found references to fltk v1.3.0 version. Looks like Dave W1HKJ did some changes about 1.5 years ago to enable compatibility to this later version. Can you update your fltk version to v1.3.0 and try building again?

73
Mauri
Logged
AG1LE
Member

Posts: 119


WWW

Ignore
« Reply #24 on: February 24, 2014, 07:16:33 PM »

Hi Mauri - just to let you know I've built your version (on Mint 16) and have successfully decoded. On first impressions it certainly seems better than the previous built-in decoder for FLDIGI. in a couple of weeks there is the UK Commonwealth contest (BERU) which is CW and restricted to UK and our commonwealth countries so it is a fairly relaxed contest with no huge pile-ups since Europe and the USA can't enter. I hope to enter this year with CW treated purely as any other data mode on FLDIGI using your decoder. Last year I cobbled together a solution using WD6CBF's CW Decoder on Windows with a logging program since WD6CNF's decoder seemed the best in tests.

Having a decent decoding engine inside FLDIGI will, of course, make this so much easier this year. Should be interesting.

Thanks for the huge amount of work that must have gone into this project and should you find any bugs or improvements then I'd love be on the test list.

Brian G0UKB


Thanks Brian

I will keep you posted once I get the next set of bug fixes and improvements done. Speed estimation algorithm needs some work, as the decoder often misses first character after a long pause.

I have an experimental standalone version where I have also tested expanding trellis path length from current 25 up to 100.  This requires much more CPU capacity, even my Intel I7 at 3.4 GHz takes significantly longer to decode. Results especially in noisy signals exceeded my expectations.  This Bayesian algorithm is the most promising so far, and I have tested quite a few different approaches already.

Good luck with the contest and please share your experiences with this decoder, both good and bad.

73
Mauri AG1LE
Logged
N3HEE
Member

Posts: 113




Ignore
« Reply #25 on: February 25, 2014, 01:41:24 AM »

Mauri, I updated to fltk1.3 and was able to build.  However now getting audio error.  E: trx_trx_receive_loop: Sound error: No such file or directory  I see that OSS is the only choice for soundcard.  My system uses portaudio.  -Joe
« Last Edit: February 25, 2014, 01:51:24 AM by KB3KJS » Logged
N3HEE
Member

Posts: 113




Ignore
« Reply #26 on: February 25, 2014, 04:33:27 AM »

Ok, got portaudio support to compile with portaudio19 dev library installed.  Decodes ARRL code practice file flawlessly.  I will do some on air testing tonight and report back then.  -Joe
« Last Edit: February 25, 2014, 04:47:40 AM by KB3KJS » Logged
N3HEE
Member

Posts: 113




Ignore
« Reply #27 on: February 25, 2014, 09:44:34 AM »

Mauri, I did some testing today.  I have to say the legacy decoder did a better job.  Are there any settings I have to adjust for optimum bayes decoding? I was just checking the bayesian filter box on the configuration page. I kept all the other settings and levels the same during the test.  I will continue to test more later.  -Joe
Logged
AG1LE
Member

Posts: 119


WWW

Ignore
« Reply #28 on: February 25, 2014, 08:07:16 PM »

Mauri, I did some testing today.  I have to say the legacy decoder did a better job.  Are there any settings I have to adjust for optimum bayes decoding? I was just checking the bayesian filter box on the configuration page. I kept all the other settings and levels the same during the test.  I will continue to test more later.  -Joe
Hi Joe

Try to keep the signal volume level relatively low.  The waterfall background should be blue. If the volume is so high that background turns yellow the decoder will have problems. Otherwise the default settings  (matched filter checked) should work fine.

73
Mauri AG1LE
Logged
N3HEE
Member

Posts: 113




Ignore
« Reply #29 on: February 26, 2014, 09:31:39 AM »

Mauri, My levels were pretty low as you described.  Sorry but the legacy decoder worked better in all cases.  Honestly though,  the legacy decoder doesn't work all that well either compared to others I have used.  How much of the legacy decoder code are you using?  I will keep an eye on your development.  I think it's pretty interesting work.  Thanks -Joe
Logged
Pages: Prev 1 [2] 3 Next   Go Up
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!