Monday, September 06, 2010

Open digital voice codec released

David Rowe has released an early version of his low bit rate open source digital voice codec. Whilst this is certainly a significant achievement, I can't get very excited about it for two reasons. First, it is incompatible with the AMBE codec used by Icom in D-Star, so it is never going to be able to work with existing D-Star equipment and it is not all that likely that it will be incorporated in future versions. And second, I still don't see what benefits digital voice offers us as radio hams in the real world.

Digital audio gives you clear "fully quieting" audio up until the signal starts to be lost, when it quickly breaks up and you get nothing. Analogue audio experiences a gradual degradation as the signal gets weaker. Those who have used the two report that an analogue signal is usably copyable about 15 - 20% further out than a digital signal. I can understand that for some professional services nothing but clear copy will do. But we are amateurs who are supposed to be good at digging weak signals out of the noise. Is this really "progress", in the context of what ham radio is used for?

The other reason for my antipathy to digital voice is that I question whether we need another voice mode at a time when VHF analogue voice operation itself seems to be in decline. Surely it would be better to be making more use of existing technologies like Echolink to create more activity for users of existing equipment than to introduce new digital modes that will have an even smaller number of users. I've seen D-Star radios up for sale because the original purchaser found there was hardly anyone to talk to, and D-Star has now been around for several years. I don't see the availability of an open source digital voice codec changing that situation somehow.

11 comments:

Theodore said...

Dstar is a zombie mode - it's dead, but does not realise it yet.
AMBE compatibility is no more an issue than the brand of shoes that the Dstar zombie is wearing.
The point of having open source codecs for voice is that it can form the nucleus for any number of ham inspired projects.
Limiting our thinking to life support for a dead mode is setting too low a bar for this codec.
There are so many talented people in ham radio that I have no doubts about the codec being used in projects which are still but vague stirrings in the creative universe.
As regards analog versus digital voice, the main drag factor has been Dstar itself, it bears the seeds of its own destruction.
But having a non-proprietory codec means uncaging digital voice development.
As digital data modes have shown, you can achieve results far down in the noise where CW and voice can only dream of communicating.
Since a digital voice codec is just producing a data stream, one could imagine integrating it with some innovative digital transport stream for truly spectacular low signal performance.
Remember, progress always comes from individual creativity - companies only create mission statements.

James said...

Sure, it isn't compatible with AMBE so you won't be able to talk to someone who can only talk AMBE2020. But some radios had the codec on a card so that can be replaced (or, I guess, supplemented somehow so you can switch between codecs). The underlying transport won't be affected by that though. I could see it being possible to talk to other codec2 users on a standard dstar repeater or someone could set a repeater up with an AMBE chip such that it would translate on the outputs I guess. Though that would probably require someone set up two output frequencies, one for AMBE and one for codec2. I don't honestly know how feasible that is now that I've mentioned it.

I think dstar is pretty neat because it can do callsign squelch and apparently you can route over the internet to talk to remote repeaters or to call a specific person by callsign where ever they last checked in from.

Julian said...

Theodore: I'm skeptical of these claims that digital voice will enable copy right down in the noise. Current digital codecs can't even equal analogue in this respect. Digital data modes only achieve it when using very low bit rates, far too low to carry voice information. This open codec is currently quite a bit inferior in performance to the AMBE one used in D-Star, though it is an early technology preview and will no doubt get better.

James: I don't think callsign squelch is neat at all though to be sure it will be a boon on all those cliquey repeaters where the users won't even have to hear the people they don't want to reply to. Echolink already allows you to connect to remote repeaters, and as for calling someone whereever they are, I already have a device that can do that called a cellphone.

James said...

Julian, This is where we differ I'm afraid. To be sure, I do support maintaining the spirit of the last 100 years of amateur radio. Don't get me wrong there. I just feel that if the service is going to stick around for another 100 years then it has to advance or become irrelevant. APRS is indeed one mode that is pretty neat. Echolink is pretty cool also, but the only digital side is on the Internet. Cellphones are exactly what I'm talking about. They have all sorts of technology that we are only just now getting or we can only dream about. In less than half the size of a standard HT you have a device that can take pictures and transmit them (SSTV), remotely track your location (APRS), send and receive text messages (PACKET, RTTY, etc), and talk around the world.
What do we have? To have the same functionality would take up a whole suitcase of radio gear, right now. I'm none-to-fond of the crowd who sees amateur radio as solely important when it comes to emergency communications, but imagine how those capabilities in an HT sized package could help.

Not trying to pick a fight, just want to share my viewpoint. I've been reading your blog for awhile and enjoy seeing new posts.

Julian said...

Hi James. Thank you for sharing your viewpoint. I think we'll have to agree to differ. I'd just like to make it clear I'm not against change because I'm against change. I just don't see the attraction of ham radio if it simply mimics the capabilities of a cellphone. I think for ham radio to survive it needs to offer something unique, something worth going to the trouble of getting a license for. If I thought digital voice was it then I'd be in favour of it. But the take-up of D-Star, which can do quite a few of the things you suggested already, has been pretty slow even though it's been around a few years now, which suggests that it isn't just me who don't think it's the new "killer app" for ham radio.

Perhaps most people just hate the idea of having to buy a rig from Icom. Certainly, having a well-sited D-Star repeater (with its built in DPRS support) would solve the problems with limited APRS coverage we have here, so in being opposed to it I'm making things more difficult in that regard.

But no new technology is worth having unless enough people are willing to use it. There are plenty of great but little used digital data modes that illustrate that point while the majority still use PSK31 and RTTY. And in that case there isn't even any expense involved in using the new technology.

Show me the evidence that hams really want to go digital and I'll be willing to agree I've been wrong about it.

Dave said...

Julian,

I think the death of DStar is all about the price point. If you could buy a DStar HT for $150, everyone would be buying into it.

Since the minimum purchase price is more like $450, you see a much much smaller group even consider the idea of buying it.

James said...

Dave, I know that's a big sticking point for me. If Wouxun or one of the other Chinese manufacturers brought out a HT that was dstar compatible or dstar with codec2 at a fairly low price point I'd be willing to bet they'd have a winner on their hands.

Julian said...

I don't know if you would ever see a D-Star radio that cheap. Even the DV-Dongles are $200, presumably due to the cost of the codec. A D-Star radio needs to have better frequency stability than an analogue FM radio so a higher standard of engineering and build quality is required, which obviously adds to the cost.

Looking at the price of D-Star radios now, they do not seem that much more than the high end dual banders from the other manufacturers. But they make it up on the accessories. A drop in charger for one of the handhelds is £60. And the GPS mic which I would certainly want it I was ever to go D-Star is £230!! You could buy three Wouxun radios for that!! I thought the GPS option for the VX-8R was expensive but this is just "'avin a larf." I don't see how they can possibly justify that price when you can buy a GPS receiver module for about £10.

Julian said...

By the way Theodore, returning to your point about digital modes allowing communication down into the noise, I was reading a discussion between some people who have tried DV on HF who are interested in trying D-Star on 10m. Their experience has been that digital breaks up very quickly when there is multi-path reflection such as is generally experienced on the HF bands. No doubt modulation methods could be devised that are resistant to that, but they might require more bandwidth or even techniques like spread spectrum that are not permitted on HF.

Of course at the moment this is all just an experiment, which is very interesting for those clever enough to do it and the rest of us to watch from the sidelines. But I don't expect to see DV taking over from analogue SSB as the mode for long distance HF communications in my lifetime at least.

James said...

I've wondered if the DV-Dongles aren't extremely overpriced either. I've recently heard that it's possible to buy the AMBE2020 chips in single quantities for $25. I don't know if the DV-Dongle does something else that runs up the price. I know it plugs into a usb port but if it's just a usb-serial link to the AMBE chip then I could cobble the same thing together for $40.

Of course, we do live in capitalist societies so the people making this stuff has an interest in charging what the market will bear.

Wow, I just looked up the price of a new IC92AD on universal radio. $570! For a dual-band HT that just happens to do digital. The speaker-mic with GPS is another $220 (Discounted $100).

That's a little excessive in my opinion but apparently people buy them otherwise the price would come down or they'd stop selling them.

Julian said...

That is interesting, James. Over here the IC92ED (European version) is £380. There is also an E80D which is £340. That's close to the prive if Yaesu's VX-8 radios. Normally the price of ham gear over here is more of a £1 = $1 conversion and I'm sure that D-Star HT's were over £400 at one time. So it would seem that perhaps Icom over here is cutting the price of D-Star radios to try to boost sales, and making it up on the accessories.