DSD 1.4 and mbelib 1.2.3 released

[MSM_Maria]

Eddy... No, last record also is bad. The data is not detected. Of course, I see what it is MotoTRBO, but it is(such a records) impossible to decode. You need to disable all the features of a sound card, e.g. any automated systems or DSP processors and etc. It is also possible the problem is RX, but probably in the process of recording.

 

[MSM_Maria]

Sorry for my English in my previous answers, i just had read my previous replys and i realized that
these demand edition, most likely because of my English You didn't understand me.

There are some problems in your records. The level of records is low, there is also small shift of polarity to positive, and there are distortions in your last record.

You need to turn off all the features in sound card and increase the level of record. The data is not detected in your last record, as the record is strongly distorted and has low level in general.

 

[hsuanwang]

Dear Maria...

Finally I find the problem... The RF signal input is not strong enough ! Of course there are some hardware parameters can be tuned to improve the DSD output quality but the RF signal is most impotent. Now I can use DSD to decode the P25 and DMR signal from our fire department. Anyway, thanks a lot !



Stupid Eddy...

 

[primehifi]

Sound Recorder works, as I said, and the DSD scope seems to recognize at least that there is something attached to the line in on the netbook.

First one is cable disconnected.. second image is the scanner idle.. third is the scanner actively receiving ProVoice.

Jay, just want to clear this up, I know it's been a while but RNC are not ProVoice, they're AEGIS Digital. They're noted as being ProVoice in the DB but that's an error that no one's ever bothered to correct.

Sadly, AEGIS will not decode in DSD. There is some info here: Digital radio with vocoding encrypting codec - Ericsson GE Mobile Communications Inc. but I'm uncertain of which information in particular is needed to demodulate the AEGIS modulation.

 

[Jay911]

Jay, just want to clear this up, I know it's been a while but RNC are not ProVoice, they're AEGIS Digital. They're noted as being ProVoice in the DB but that's an error that no one's ever bothered to correct.

Has it ever been submitted?

 

[primehifi]

Has it ever been submitted?

You've been the only one submitting info to the DB for NL. From what I've heard from friends (self included) is that it's a little too cumbersome to submit info to the db, so no one bothers when really there isn't a whole lot to scan here anyway.

But yeah. We compared the AEGIS from RNC with a bunch of provoice samples and determined that the RNC is Aegis. The confusion stems from the fact that A) few know the real difference of AEGIS and ProVoice to begin with and B) a lot of the sample audio clips online are again, mixed up. Sites calling ProVoice, AEGIS and AEGIS ProVoice. It was quite the pain in sorting it out.

There were other elements that went into this confirmation but I won't get into them. Based on some other findings, it's assumed all their traffic is ENC as well :-/ but without having a M-RK that does VHF, AEGIS and EDACS trunking there is no way to know. Quite hard to find.

The final nail in the coffin is that the homepatrol will show if a tx is digital or analog and all the RNC traffic is digital.

Locally, it was thought for quite some time that it was AEGIS and not ProVoice but like I said, no one ever cared enough to correct the DB as up until about 2 weeks ago it wasn't 100% confirmed, just assumed.

 

[kd4efm]

DSD and mbelib on CentOS 5

I can say I can run DSD on CentOS 5, I can get speratic decodes from a P25-ph1 but not reliably enough...
Can not decode NXDN right now, radio does not go where I need it to go...

Centos 5 2.6.18.274
P4 1.7Ghz
Onboard sound
using Line In @ +/- 60%

Radio: ICOM 2820 with D-Star installed,
audio out cord, using the 9600 audio port (DISCRM) (*works with ATCSMonitor so cable is good)

P25 phase 1 system; hicups and burps.... some words decoded...
Next test, TRBO
Conv_P25 not sure what the issue is there...
more testing going on....

 

[Jay911]

I've just installed Commodore VisionOS (yes, that Commodore) on one of my older laptops as a trial. If I have time tomorrow it's my intent to put DSD on it (COS is a version of Linux with some extras) and see if it works. I'll get back and post the results when I get it done.

 

[mwaldron]

Raspberry Pi

I was wondering what the processing power requirements for DSD were, specifically I was wondering about the possibility of using the newly developed Raspberry Pi (Raspberry Pi | An ARM GNU/Linux box for $25. Take a byte!) Linux appliance to run DSD in a a black box fashion.

I know that a USB audio card would need to be added for input (the Pi has built in output but requires USB class audio for input), and the BCM2835 seems to have a fair amount of AV processing power.

I was thinking along the lines of a headless "discriminator in, demodulated audio out" appliance type build.

I see in the Wiki that Debian has worked on x86 hardware, and the ARM variant of Debian (along with Fedora and ArchLinux) are supported OS' for the Raspberry.

Anyone else consider this? Is it a pipe dream? Did I ask in the wrong place?

 

[studgeman]

in a quick look through the code, i didn't see any routines that were x86 specific. You should be able to cross-compile for ARM by setting the right flags for GCC. Give it a shot. Most of the processing for DSD is done in the soundcard DSP, the actual CPU requirements are low for modern processors. You may run into some issues with soundcard drivers, choose wisely

 

[Byrus]

I've tried for the 3 days to get a version of ubuntu to work with my sound card. Finally got sound working, but when I type dsd this is what it shows
Digital Speech Decoder 1.4.1
mbelib version 1.2.3
Audio In/Out Device: /dev/audio.
Is that suppose to happen. No errors, but nothing else.

 

[ilgrant]

Byrus-

That is what is supposed to happen when there is no digital signal to decode. If you are sending in one of the supported signals and there is nothing being displayed on the screen then there is still a problem somewhere.

 

[Byrus]

Thank you. I think the p25 signal is to weak.

 

[Byrus]

I've picked up a nxdn96 signal, but I think my pc is running to fast it shows voice E= then scrolls to the next line real fast and shows data. I hear garlbed voices, but they sound like fast fwd mode.

Sent from my SGH-I897 using Tapatalk

 

[ilgrant]

The speed of your computer shouldn't be an issue, Byrus. What you described has been my experience as well. Someone mentioned earlier that it may not handle trunked nxdn correctly which is what I get around here.

 

[kb1ipd]

Are there any updates on the progress of decoding Opensky? It would seem that Opensky has been the single hardest thing to tackle. The last I heard was that decoding of Opensky with DSD was in development but that the developers were in need for some more raw audio samples. I'm sure there must be someone on here who can provide those. Pretty much anyone living in NY or PA should have the ability to receive some traffic and record the raw audio.

There was also some concern about whether Opensky required encryption. I have no idea if that question has been resolved either.

So... any progress to report on that front?

 

[mtindor]

Are there any updates on the progress of decoding Opensky? It would seem that Opensky has been the single hardest thing to tackle. The last I heard was that decoding of Opensky with DSD was in development but that the developers were in need for some more raw audio samples. I'm sure there must be someone on here who can provide those. Pretty much anyone living in NY or PA should have the ability to receive some traffic and record the raw audio.

There was also some concern about whether Opensky required encryption. I have no idea if that question has been resolved either.

So... any progress to report on that front?

dsdauthor hasn't posted here since Oct 2010. There haven't been any updates in more than a year to DSD. My guess is either the big radio manufacturers did something to curtail his activities, or they hired him rofl.

At any rate, looks like not much of anything is happening with it.... unless some people have taken the source code and are working behinds the scenes to incorporate it into other applications, etc. [I can think of at least one person who might be doing such a thing].

For now, we're stuck with what we have.

M

 

[kb1ipd]

dsdauthor hasn't posted here since Oct 2010. There haven't been any updates in more than a year to DSD. My guess is either the big radio manufacturers did something to curtail his activities, or they hired him rofl.

At any rate, looks like not much of anything is happening with it.... unless some people have taken the source code and are working behinds the scenes to incorporate it into other applications, etc. [I can think of at least one person who might be doing such a thing].

For now, we're stuck with what we have.

M

Or he (or she) just got busy with other things and has not had the time to maintain a project like this. This often happens with open source software, especially when it's one person or a couple people doing all the developing.

It would be nice to hear about what the situation is, though. If the author of DSD is no longer to do it, then that's fine (s)he has already contributed a massive amount to this hobby and nobody could fault them for not being able to go further.

It would just be good to know so it could be determined whether someone else who can do that kind of programing needs to step up and take over the project.

 

[MattSR]

My guess is either the big radio manufacturers did something to curtail his activities, or they hired him rofl.

I'd bet money that it was neither. He's a self motivated hacker - I'd bet he simply got bored with the project

 

[Nessus]

Tried the matching filter

Hello KA1RBI

Maybe You are wrong, and proper value is 0.2 or very near about this.

In general, if You know what type of filter RRC, You can get the alpha value is very close to the truth.

Please look images.

A good visible NXDN has alpha not 0.1 but very really alpha = 0.2 or near.

We write small article about problem with output from discrimanator and program decoding/pocessing An output from discriminator and program processing. I hope it was not in vain.

Hi Maria,

That is a great little tool that you have there. Regarding the article, I think you folks are saying that addign an RRC type matching filter at the input to the demodulator will narrow the symbol histogram. OK, I'll buy that. So I modified dsd to incorporate a matching filter for C4FM as specified in TIA-102. As you may recall from the CAI spec, the D() filter frequency response specification is

|D(f)| = sin(f*pi()/4800)/(f*pi()/4800) for the frequency range |f| < 2880 Hz

Base on this, I have constructed a symmetric FIR with the following impulse response (assuming a sampling rate of 48kHz):

#define dfiltlen 26
#define scale 1.0

static const double d_filt[dfiltlen/2] = { /* b0 b1 b2 ... b(NH-1) */
/* Note: Symmetric FIR so we only list one half of the coefficients */
/* Gotta be even obviously */
scale * -0.091815186121074741
,scale * 0.0064564715437959441
,scale * 0.010924280987368085
,scale * 0.018047303559653588
,scale * 0.027451231863106759
,scale * 0.038526242683991185
,scale * 0.050660314980775376
,scale * 0.063022769134530962
,scale * 0.074957923169934604
,scale * 0.08575818347014047
,scale * 0.09427717139504066
,scale * 0.10037174500934753
,scale * 0.10351765408516761
};


I have experimented with various C4FM inputs in dsd with and without this filter. I do not see a significant difference in the Histogram base width.

If I hade the windows build environment, I could publish executables for you to try this with.

Cheers,