bad audio quality of D-Star on DSD1.70

[ronenp]

Hi there
Recently we start to experiment D-star
When the D-star is decoded with the DSD the audio quality is much less then the D-star radio decode
Decoding DMR and P25 with the DSD gives much closer decode quality compered to a real radio
Can something be done ?
Please advice
Thanks Forward
Ronen - 4Z4ZQ
Ronen Pinchooks (4Z4ZQ) WebSite

 

[kb9mwr]

Hi Ronen,

What I learned from Bruce Perens AMBE Exposed talk is that APCO Project 25 traces back to the early 1990's and required publication of the IMBE and AMBE codec algorithms, which of course is how people figured out how to implement it in DSD.

D-Star is a whole different animal. It uses an earlier version of AMBE, that wasn't documented anywhere, so that is likely why it doesn't sound as good.

Here are the more technical details of various folks who were working on it a few years ago, and were able to bring us what we have today in DSD for D-Star support:

http://forums.radioreference.com/vo...tware/215282-decoding-d-star-any-success.html

 

[groovy]

I echo what kb9mwr said- The Dstar codec is completely undocumented, and the work on it was possible through firmware reverse engineering and guess-and-test with AMBE2020 chips. I think the developers who were working on it made their point, and then moved on to other things- I haven't seen any commits to the dstar related files for some time.

If anybody wants to continue the research / work, I suggest you look at the osmocom GMR code that Sylvain Munaut worked on. Those phones use a similar codec - I believe with longer frames for the satellite latency. Initially he used the mbelib code, enhanced it for things like tone support, but he later rewrote the synthesis code completely. See OsmocomGMR for his presentations and source code.

Unless other developers take this on, DVSI releases specs, or you switch to a hardware-based decoder like the thumbDV, I don't see the dstar voice quality improving in the short term.

 

[ronenp]

Isnt the D-star supposed to be open code and as part of it the AMBE specifications ?

Thanks Forward
Ronen - 4Z4ZQ
Ronen Pinchooks (4Z4ZQ) WebSite

 

[KD8DVR]

Isnt the D-star supposed to be open code and as part of it the AMBE specifications ?

Thanks Forward
Ronen - 4Z4ZQ
Ronen Pinchooks (4Z4ZQ) WebSite

DVSI owns(ed) the patents on AMBE. Definitely closed source. Word is, however, the patents expired a month ago. I've really heard of no one in the digital voice forums of anyone trying to exploit this yet. I'd assume there would be folks all over it trying to work on a chip-less D-STAR.

 

[kb9mwr]

The AMBE chip has the read fuses blown off, which makes reading it impossible. Else that would probably be the easiest way to understand how its done.

ADCR25 a stand alone P25/DMR/Fusion/NXDN receiver

There is no AMBE chip in this implemenation (done in DSP)

ADCR25 a stand alone P25/DMR/Fusion/NXDN receiver

He notes:
FT-1DR consists of DSP unit board based on TMS320 DSP not DVSI chip.

He have heard similar reports on some of the newer D-Star radios. Maybe someone can verify models?

As for folks being all over a chipless D-Star, I think people with that expertise are few and far between. Couple that with an unclear picture of any possible legal challenges and you have even fewer. Kudos to those who brought us where we are today,