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: [1]   Go Down
  Print  
Author Topic: Multichannel CW decoder for FLDIGI  (Read 1941 times)
AG1LE
Member

Posts: 114


WWW

Ignore
« on: July 17, 2014, 06:16:12 PM »

I have been working on the Bayesian Morse decoder for a while.  The latest effort was focused on making it possible to automatically detect
all CW signals in the audio band and spin up a new instance of the Bayesian decoder for each detected signal.

See the blog about experimental version of FLDIGI with this feature:
http://ag1le.blogspot.com/2014/07/new-morse-decoder-part-5.html



I have saved the full sources of this experimental FLDIGI version in Github:https://github.com/ag1le/morse-wip/blob/master/test/fldigi-3.21.77.multich.tar.gz

Let me know if you are interested in testing this software.  I would be interested in getting feedback on scalability, any performance problems as well as how well the Bayesian CW decoder works with real life signals.

73
Mauri AG1LE
Logged
AG1LE
Member

Posts: 114


WWW

Ignore
« Reply #1 on: July 19, 2014, 03:26:41 AM »

It looks like most people would like to have pre-compiled Windows version in order to test this. Let me check how to make that available.

Some early comments posted in here:
http://ag1le.blogspot.com/2014/07/new-morse-decoder-part-5.html?showComment=1405761369426
Logged
N0SYA
Member

Posts: 309




Ignore
« Reply #2 on: July 19, 2014, 06:36:37 AM »

I'll provide some feddback once you get a windows ver up.
Logged

If you have a clumsy child, you make them wear a helmet. If you have death prone children, you keep a few clones of them in your lab.
ZS6DX
Member

Posts: 13


WWW

Ignore
« Reply #3 on: July 20, 2014, 07:32:17 AM »

I have been working on the Bayesian Morse decoder for a while.  The latest effort was focused on making it possible to automatically detect
all CW signals in the audio band and spin up a new instance of the Bayesian decoder for each detected signal.

See the blog about experimental version of FLDIGI with this feature:
http://ag1le.blogspot.com/2014/07/new-morse-decoder-part-5.html



I have saved the full sources of this experimental FLDIGI version in Github:https://github.com/ag1le/morse-wip/blob/master/test/fldigi-3.21.77.multich.tar.gz

Let me know if you are interested in testing this software.  I would be interested in getting feedback on scalability, any performance problems as well as how well the Bayesian CW decoder works with real life signals.

73
Mauri AG1LE
This sounds really interesting and useful Mauri. I would love to test it!

Thanks for your hard work,
73, Rudi de ZS6DX/V51VE
Logged
PA0WV
Member

Posts: 94




Ignore
« Reply #4 on: July 20, 2014, 09:10:32 AM »

On http://pa0wv.home.xs4all.nl/TIMorsedecoder/TIMosedecoder.html

there are 5 audio files (.wav) playable under windows. They are only abt 6 second duration.
So you can play them with Microsoft Mediaplayer on repeat when you like for measurements and the like.
Morse code in pretty white noise with different signat to noise ratio's.

It is interesting to know which file numbers you are able to copy with your method under development.
Logged
AG1LE
Member

Posts: 114


WWW

Ignore
« Reply #5 on: July 20, 2014, 06:40:56 PM »

On http://pa0wv.home.xs4all.nl/TIMorsedecoder/TIMosedecoder.html

there are 5 audio files (.wav) playable under windows. They are only abt 6 second duration.
So you can play them with Microsoft Mediaplayer on repeat when you like for measurements and the like.
Morse code in pretty white noise with different signat to noise ratio's.

It is interesting to know which file numbers you are able to copy with your method under development.
Hi Wim
I did a quick test and timorse3.wav decodes cleanly. timorse2.wav decodes "TTST LE PA0WV" with FFT filter bandwidth 22 Hz.
I was not able to get anything meaningful from timorse1.wav.  These were done using the legacy FLDIGI SOM decoder.

Bayesian decoder didn't do as well. I think the current front end signal processing chain in FLDIGI is not optimal for the Bayesian decoder as noted I my earlier testing, see http://ag1le.blogspot.com/2014/06/new-morse-decoder-part-4.html.
The graph shows CER vs. SNR  at various speed (which automatically adjusts filter BW) settings


My current focus is to get the multichannel CW decoder into decent shape and then return to the problem how to optimize CER vs. SNR  vs. automatic speed detection vs.  filter bandwidth settings etc.  Thanks for very good test material, these are really good benchmark files.

73
Mauri AG1LE
Logged
PA0WV
Member

Posts: 94




Ignore
« Reply #6 on: July 21, 2014, 12:19:53 AM »

That is an excellent result Mauri!

A conventional detection with a touch tone decoder phase locked loop can't detect file 3 flawless. File 4 will do.

So your decoding method is an order better then the conventional one.  That is really a bench mark. Congrats with these results.
Logged
Pages: [1]   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!