CW Decode from HO-68 using MultimonNG - Update II

(This is an update to this post)

I tweaked the string that my stupid search script uses to be "BJ1SA " - this should then take into account the proper spacing of the characters. The best result I got out from that was this (I used the data to which I'd applied a narrow band filter):

 J1RA XW XW AAA TTT AW- ATE ETT TTT TTT TTT TTT TTT TTT TTT TTT XW XW EE
BJ1SA XW XW AAA TTT AU4 TBE ETT TTT TTT TTT TTT TTT TTT TTT TTT XW XW
BJ1SA XW XW AAA TTT AU4 TNT ETT TTT TTT TTT TTT TTT TTT TTT TTT XW XW
BJ1SA XW XW AAA TTT AU4 AVT ETT TTT TTT TTT TTT TTT TTT TTT TTT XW XW
BJ1SA XW XW AAU T5H H 5 <._.......>DHSH IIIEIS

BJ1SA XW XW AAA TTT AU4 AVE ETT TTT TTT TTT TTT TTT TTT TTT TTT XW XW
BJ1SA XW XW AAA TTT AU4 TBE ETT TTT TTT TTT TTT TTT TTT TTT TTT XW XW E
BJ1SA XW XW AAA TTT AU4 AVT ETT TTT TTT TTT TTT TTT TTT TTT TTT XW XW EIE
BJ1SA XW XW AAA TTT AW4 TDT ENTTTT TTTETTNTTTTTNAAAVATUA<....._>

This appears to be better than the previous decoding. The graphical representation of the search space is show here:

As you can see there is a much smaller range of parameters where the best decoding occurs.

I have now integrated in the multimonNG decoding of CW signals into my GroundStation software and have decoded signals in realtime from a number of satellites including FO-29 and CO-55. The current implementation I use is far from optimum as I have around 2.5kHz of audio going into multimonNG, so as mentioned previously I'll be able improve the performance by additional filtering.

I've also improved the calibration of one of my RTLSDR receivers such that I can receive meteor signals from the GRAVES transmitter - more data to follow on that subject at a later date.