DSDPlus DSD 1.8.3 & DSD 1.8.4 do not decode the voice of NXDN48 and NXDN96.

[KrisMar]

Hi. Many thanks to colleague Louis for a great contribution to the program and development of DSD 1.7 project of colleague Jared Szecha. Unfortunately I noticed that the DSD of Louis since version 1.8.3 does not decode the voice on NXDN48 and NXDN96 is only gibberish. The older version 1.8.2 still decoded well and the voice was clear and crisp. Does anyone still have this problem and does anyone know how to solve this problem so that the latest version correctly decodes NXDN ?

This version is the last to decode voice on NXDN48 and NXDN96.

https://github.com/LouisErigHerve/dsd/tree/7d3b4399531dee59e7219f04b685e9f55e431aed

And this next version no longer does. And all upwards no longer decode the pure voice of NXDN48 and NXDN96.

https://github.com/LouisErigHerve/dsd/tree/b01dc12ee1d2aea5ae2fb50466cf4278e4aadd91

Below is a link to a description of the changes.

Commits · LouisErigHerve/dsd

Digital Speech Decoder. Contribute to LouisErigHerve/dsd development by creating an account on GitHub.

github.com

 

[noamlivne]

I wonder if you inform him in his Github page, will he respond?

 

[KrisMar]

I don't have a Github account unfortunately. I wrote here a topic about this problem because I think that maybe my colleague Louis is also sometimes here, will read this topic and maybe he will fix NXDN decoding in dsd 1.8.4.

 

[lwvmobile]

He is already aware of the NXDN issues.
And I quote...

DSD is unstable when decoding trunked NXDN networks, I've tried to fix it several time with no success...


NXDN detection require a deep work to become more stable. One main problem is the SYNC pattern detection (FSW + LICH) : as these one are very short and contains only 1 CRC bit there are many false positive and these one also generate a lot of noise.


The FSW + LICH detection is used to set the FSK symbol decoder parameters, so any false positive detection will set bad parameters...


Any help would be welcomed...

LouisErigHerve dsd · Discussions

Explore the GitHub Discussions forum for LouisErigHerve dsd. Discuss code, ask questions & collaborate with the developer community.

github.com

The general takeaway I had from looking up this exact issue before was that he attempted to do some work on it, and made it worse. I could have sworn the github page had this under the issues tab at one point, but maybe it was all under discussions instead.

 

[pi9ajn]

And this next version no longer does. And all upwards no longer decode the pure voice of NXDN48 and NXDN96.

https://github.com/LouisErigHerve/dsd/tree/b01dc12ee1d2aea5ae2fb50466cf4278e4aadd91

Version 1.8.2 (both 32b and 64b) can still be downloaded from his website:

DSD – 1GénieSup

DSD - Digital Speech Decoder - DMR - dPMR - NXDN - APCO 25 - P25 - ProVoice - D-Star - SDR - Software Define Radio - Radio - PMR

www.1geniesup.fr

(downloads are at the bottom of the page).

 

[KrisMar]

Thanks for providing the link to the program. I know this site and I have all versions of this program because I compile them myself. I only reported the problem that occurred and from which change, so someone could fix it. I also have this question if there is a possibility of adding support for FUSION C4FM modulation in the future.

 

[KrisMar]

I also noticed that all versions of Louis do not decode well DMR received on direct, the voice is jagged and single syllables are heard. Only version 1.7 of szechyjs clearly decodes DMR received on direct and listen to a smooth conversation.

 

[KrisMar]

I noticed that the problem is the file "dmr_voice.c" . Is anyone able to fix this file so that "Direct DMR " works correctly? I mean to correct file from this version "dsd 1.8.4", not copy file from version e.g. "dsd 1.7".

 

[lwvmobile]

I've addressed the NXDN sync detection and voice decoding woes in my own fork, I posted a thread about it here if you are interested in trying it out. @KrisMar I could use somebody like you to test it out for me, let me know how it works for you. I don't have any of the DMR stuff from the LEH fork though, but I need somebody who tests these kinds of things out to give me some feedback.

DSD FME

Thought I'd just take a moment to share a project I've begun working on lately. If anybody wanted to give it a try out and see how it works, I'd love to hear any feedback, critique, constructive criticism, ideas, etc. This version of DSD is a flavor blend of szechyjs RTL branch and some of my...

forums.radioreference.com

 

[KrisMar]

I have tested the NXDN48 and NXDN96 KENWOOD files directly and yes, the "DSD 1.7" version decodes voice on both modulations, unfortunately your "DSD FME" version does not decode voice and has problems reading this modulation, it only occasionally shows a frame and that's it . I don't know if this is an issue with the radio or ICOM transmitting the same way as KENWOOD, or if this could be the reason that some files are decoded and others are not. The "DSD 1.8.2" version decodes both versions of NXDN modulation well. The "DSD 1.8.4" version mutters and displays RAN errors. As for DMR, the "DSD 1.7" and DSD FME versions decode voice nicely both through the repeater and directly. DSD versions 1.8.0 through 1.8.4 decode DMR well through the repeater unfortunately directly the voice is distorted.

The problem is that it is hard to find samples of NXDN48, NXDN96 direct and repeater from ICOM and KENWOOD on the net.

 

[lwvmobile]

I have tested the NXDN48 and NXDN96 KENWOOD files directly and yes, the "DSD 1.7" version decodes voice on both modulations, unfortunately your "DSD FME" version does not decode voice and has problems reading this modulation

Hmmmm interesting. I've got it working quite well on the two local NXDN48 sites that I have. What's the full command you are using to run DSD FME? I removed NXDN96 from the -fa option, so you have to manually specify -fi for NXDN48, or -fn for NXDN96 when you want to use those. Also, make sure you aren't using -W for the audio monitor, that's still a work in progress/test but sometimes when its turned on, it won't decode anything.

Here's an example command I use to monitor NXDN48 with pa input and OSS output.

padsp -m NXDN48 -- ./dsd -u 10 -fi -i pa:8 -o /dev/audio -d ./LO/ 2> /dev/null

I also don't have any NXDN96 nearby, or any sample wav files from different systems either to compare it to, but I'll make you a nice solid chunk of NXDN48 wav file from my local site so you can have it for comparison, etc.

 

[peluson]

Hi, colleague Louis's link to download v1.8.2 doesn't work, do you know if they're uploaded somewhere else?

 

[lwvmobile]

Hi, colleague Louis's link to download v1.8.2 doesn't work, do you know if they're uploaded somewhere else?

Try one of these, I assume you mean you want the Windows compiled version.

http://www.1geniesup.fr/download/dsd_V01_08_02_build_2020-06-16_Cygwin_x64.zip

or

http://www.1geniesup.fr/download/dsd_V01_08_02_build_2020-06-25_Cygwin_x86.zip

 

[peluson]

Impossible to download it is deactivated, thank you very much anyway

 

[ve2mel]

peluson,

sorry , but the files still here...

i download it with no probleme

 

[peluson]

If I had a problem with the antivirus, thank you all for your help