Digital Speech Decoder and mbelib

[millam]

dsd help

dsd -h gives you the - options and the details of each. this maybe the default help screen for c+ or Linux don't know i am a newbie to each.

 

[mike_s104]

I got it working. It works OK with the expected distortion on the Martinsburg WV tower for the WVIRP. Now I need to tap a HT analog scanner and install Ubuntu on a laptop to try it out on the Loudoun system.

 

[wvscanner]

Here is what I did to install using Ununtu 9.10 w/ Gnome as the desktop environment
Sound worked out of the box and did not have to load any third party driver.

Extract mbelib into own folder.
Extract dsd1.0.3 into own folder.

cd into the mbelib folder and typed:
sudo make
sudo make install
(first time using sudo will ask for your password)

Then I went into the dsd folder and did the same
sudo make
sudo make install
I believe it installed to /usr/local/bin or something like that (I'm not in front of the computer right now).

Finally after it was installed, in the command prompt I typed in: dsd
and started picking up digital voice from the scanner I had connected to the mic input. It was pretty distorted at times, but this is likely due to a slow laptop and crappy on-board sound (which I understand doesn't work very good). I am going to try on my desktop this evening and hopefully have better luck in decode performance since it has Line-in which is most likely better than a microphone input. But as early alpha/proof of concept software, I am both very impressed and excited to see what the future brings.

Can anyone help on this.
I try with a vmware got down to type in dsd and a few lines then nothing happen. Do I run this from the directory I was working or was I pose to went to the installed to /usr/local/bin.?? I know nothing about linux so please keep it down to linux for dummies for me.
Thank You,
Jeff

 

[mike_s104]

Can anyone help on this.
I try with a vmware got down to type in dsd and a few lines then nothing happen. Do I run this from the directory I was working or was I pose to went to the installed to /usr/local/bin.?? I know nothing about linux so please keep it down to linux for dummies for me.
Thank You,
Jeff

you need to run it from the directory where it was compiled, not installed.

 

[blantonl]

Folks, I trimmed out all the off topic posts to this thread.

 

[mike_s104]

Folks, I trimmed out all the off topic posts to this thread.

Lindsay, have you tried this with the San Antonio/Bexar County EDACS ProVoice system yet?

 

[dsdauthor]

Lindsay, have you tried this with the San Antonio/Bexar County EDACS ProVoice system yet?

The README file in the source package lists the formats that are currently supported. ProVoice is not one of them. If you would like a frame format supported upload a .wav file with several minutes of voice channel activity somewhere. The format should be 48k/16bit/mono from a disc. tap. There are no guarantees that I can get it to work but that is where we start.

I finally have serial port resume scan commands working for both GRE/RS and Uniden scanners.
I plan to release an update with that tomorrow after some testing. I am also going to release
an updated mbelib with lowered error rate thresholds to reduce the noise bursts.

Is there any way for me to edit the original post so I can put links to the latest versions there?

 

[mancow]

What about the WIKI? Can updates and info be kept up to date there?

 

[mike_s104]

The README file in the source package lists the formats that are currently supported. ProVoice is not one of them.

Sorry, I was going off the Source Forge that mentioned ProVoice. I should have readme in the download.

Nevertheless, this is a step in the right direction for the Motorola Phase II systems like in Loudoun VA.

 

[blantonl]

I think the Wiki would be a great first step for documenting and providing links to the downloads.

At this time though, I'd rather RadioReference not host the actual application code itself.

Interesting project!

 

[woodpecker]

The README file in the source package lists the formats that are currently supported. ProVoice is not one of them. If you would like a frame format supported upload a .wav file with several minutes of voice channel activity somewhere. The format should be 48k/16bit/mono from a disc. tap. There are no guarantees that I can get it to work but that is where we start.

I finally have serial port resume scan commands working for both GRE/RS and Uniden scanners.
I plan to release an update with that tomorrow after some testing. I am also going to release
an updated mbelib with lowered error rate thresholds to reduce the noise bursts.

Is there any way for me to edit the original post so I can put links to the latest versions there?

Is it possible for you to add mototrbo, I can upload a wav if required? TIA

 

[KA1RBI]

1. It uses QPSK modulation so there are many locations where signals from multiple towers
make decoding from an FM discriminator difficult to impossible. You will have to find a location that gives good error rates [snip]

The LSM (simulcast distortion) problem is slowly yielding some of its secrets. The discriminator tap method has a lot of trouble with LSM/CQPSK as we well know. It now appears that we might lay some of the blame upon the final IF (normally 455 KHz) inasmuch as I'm starting to suspect that some phasing distortion (group delay) might be coming from there. Not as much of a problem on voice or C4FM, but it can be a real killer on phase-moduated sigs...

Additionally, the typical scanner is never rated on the phase-noise performance of its local oscillator(s). If the phase noise is poor (not unlikely) the result will be compromised performance on reception of phase-modulated sigs....

Finally, it may be possible to start understanding what exactly is happening when rotating the beam antenna for "Best" reception. When looking at the received signal constellation (using Gnu radio), a "normal" PSK signal has only an FM component - meaning that the signal points all appear around a circle of fixed radius. However LSM signals have an AM component which (depending on the SNR at the receiver [?]) will sometimes cause the constellation to appear as a large "plus" or "cross" shape such that instead of all symbols in a quadrant appearing at the edge (circumference) of the circle they can appear along a line or band anywhere between the (0,0) center point and the edge as before. Ideally they would all migrate out towards the edge as in a conventional constellation. As long as they are at least some distance out from the center all should be OK. However when there's a large cluster of symbols right in the center there's no way to tell which quadrant they're in. A pic is worth 1,000 words and I'm hoping to get some online illustrating the above scenarios, when I get a fresh supply of round tuits...

In the longer term I'm hopeful that solutions will appear for modifying ordinary scanners (such as using an IF-tap) but may take more time than some have patience for (if so go buy a $500 radio!)...

Best Regards

Max

 

[dsdauthor]

Here is version 1.1 of DSD and 1.0.2 of mbelib.

DSD-1.1 is able to properly monitor Loudoun County's TDMA system now that it supports scanner control via serial port. It also has much better handling of the TDMA frames. I am close to having the Link Control (LC) bits sorted out which should give talkgroup and source info in a future release.

dsd-1.1-src.tar
Megaupload Download
Sharebee Download

mbelib-1.0.2-src.tar
Megaupload Download
Sharebee Download


Scanner control was tested on pro-197 and BC-780xlt so it should work on scanners compatible with those. Resume scan is essential to monitoring systems like Loudoun that transmit data packets (PDU's) on the system almost continuously. This causes the scanner to get stuck on the non voice channels while scanning in conventional mode (which is required to get the TDMA talkgroups). use dsd -h to see all command line options.

DSD 1.1 changes:

1.1 New features:
Scanner control options to allow scan resume during certain TDULC
Improved upsampling function audio quality
Greatly improved handling of MoTDMA frames
Much faster QPSK decision point tracking by default

Fixed bugs:
playMbeFiles was not output to .wav file when -w was given
Now correctly detects/handles Mot and P25 talkgroup formats
PDU frames were not detected

mbelib 1.0.2 changes:

1.0.2 Fixed bugs:
fixed several problems in synthesizer causing clicks
reduced error -> repeat thresholds for reduced noise bursts
ambe3600x2250 now correctly handles "Silence" frames
ambe Tone and Erasure frames are now indicated with T and E in errorbars

Known issues:
There are still some minor "clicks" in the synthesized speech regardles of output sample rate or buffering configurations. AMBE speech still sounds a little weird on some transmissions. AMBE Tone frames are not synthesized yet.

 

[SCPD]

I spoke to the author of this thread and he states he does not have the skills to do this.

Can anyone port this open source over to Windows XP or windows mobile?

 

[AZScanner]

I spoke to the author of this thread and he states he does not have the skills to do this.

Can anyone port this open source over to Windows XP or windows mobile?

I wish I had the skills (I could attempt it, but I doubt it'd work). Maybe this weekend I'll have some time to futz around with it but again, I doubt I'll create anything other than a program that goes kaboom when you launch it. But WTH, Visual Studio Express is free. Maybe I'll give it a go for fun.

Unitrunker's the guy I'd put good money on to have the chops to get this done, but he's probably nervous about what DVSI would say. Maybe all we need is a helpful push. From what I understand of C++ (which is damn little, I should have paid more attention to it in the C classes I took in college dammit) it appears as though the only linux dependent pieces of this thing are dsd_audio.c, and dsd.h - the rest should port over as is. Anyone with more knowledge agree (and would maybe be willing to try it?) That way you're not violating anyone's patents since neither of those two files does any of the verboten decoding magic.

-AZ

 

[gary123]

One feature that would be nice is an additional operation mode that just displays info such as the NAC, sourceID, destinationID, TGID,Encryption type, and any other info that would be user specific.

This is a great routine with lots of potential. Its easy to install and operate. A win version would be icing on the cake. I have all the field frame info figured out and can probably help someone with some of the info about the HDU,LDU fields if they are able to create the VB or C++ code.

 

[CrazyCalifornian]

The Scope -s option shows most of that...

 

[slicerwizard]

Can anyone port this open source over to Windows XP or windows mobile?

Already being done.

 

[313RADIO]

dose the project have a dedicated website ?

 

[dsdauthor]

There is no official website at this time as the author(s) wish to remain as anonymous as possible. I did create a wiki page here at RadioReference:

RadioReference Wiki: DSD

I will keep that updated with links to the latest versions from now on.