locked Re: WSJT Sensitivity V MSHV #why ?

Derek Brown

Thanks to all who have picked up on this subject, at this point I am afraid I will declare myself out of the ongoing discussion, as I am about to move homes, and the chances of me getting back on 6M Es before next June are pretty slim, let alone any time after today, soon to have a unknown period of enfored non-operation, until I have a shack to setup in and antenna to connect to.

73 all derek G8ECI signing off from JO03 square to qsy 15km north to IO93

On Monday, 21 June 2021, 15:36:21 BST, Bill Somerville <g4wjs@...> wrote:

On 20/06/2021 17:55, Chris, HB9DFG wrote:
Hello all

With interest I have read through this thread and I thought to myself: why shouldn't I test this as well with JTMS, since I assumed that all other
programmes concerning the FT8 decoder should actually react in the same way. I once compared an older Version of MSHV but it lost a lot
of decodes compared with WSJT-X so I did not tried anew since then.

Well, I now have tested JTDX 2.2.156 and WSJT-X 2.4.0c19d62 a short time (for about 30-40min) on a quiet 2m-band. The computer is still a
WIN7-machine with 8GB RAM and an onboard Realtek audiocard set to the 16bit 48000kHz DVD system. BKTIMESYNC is in use here als well.

WSJT-Xs decoder was set to "deep" and "AP", JTDXs decoder was set to: threats=auto, decoder cycles=1, decoder sensitivity=use low thesholds

Besides the good decodes on both programs (with big differences in the dB values), I took out 13 wave files where either WSJT-X or JTDX did not
decode something.


- All in all I have picked one periode where WSJT-X had 3 decodes (include one a1), and FTDX got 2 of them. Both programmes were able to read the
   corresponding other wave files of this periode where JTDX couldn't read the "a priori" decode.
- An other periode got 2 decodes on JTDX and one on WSJT-X
- Also one periode got 1 decode on WSJT-X and no decode on JTDX
- The rest of the periods (10) got 1 decode on JTDX and no decode on WSJT-X
- WSJT-X could read JTDXs only no-decode
- JTDX could read 5 of WSJT-Xs 10 no-decodes

Maybe if time permits I will try to make an other test with my other W10-computer with the "soundcard" of my IC-9700.

Bill Somerville wrote earlier:
"...the only way to analyze the issue you report is for you to provide a .WAV file recorded by WSJT-X along with a list of the messages that you think
should have been decoded. Anything else is simply subjective speculation."

Fine Bill, did you already got such files to investigate and did you found out something interesting? Or would you like to get my wav-files to investigate?
If yes, where can I sent them as zipped files?

Best regards de
Chris, HB9DFG

Hi Chris,

thanks for doing this objective analysis work, the examples would be very useful. Do you have any free Cloud storage accounts like Dropbox, Google Drive, One Drive, etc. where you can upload the files and provide a download link?


Join main@WSJTX.groups.io to automatically receive all group messages.