Graves on SOHO Technology

I don’t generally hang around in public VoIP forums but I do monitor several using Google Reader. So it was that Tuesday evening I found a new thread about “HDVoice” at the VoIP Tech Chat forum on BroadbandReports.com. Someone had posted a question about the significance of HDVoice and a few braves souls were weighing in with their take on the matter.

There was some good information, but a little confusion about data rates and the possible use of ATAs with wideband voice. It looked like the start of a good conversation on the subject, in a potentially interesting crowd. I thought I’d offer my two cents…which kind of unexpectedly turned into more like half a dollar.

Here’s the original thead, and now my contribution.

I now feel compelled to weight in here since wideband voice is a very dear subject to me, and one that I often cover both on VUC calls and my blog.

Firstly, G.722 is simply the most commonly available wideband codec. To be more specific, it appears in a significant number of hard phones, several soft phones and many conferencing systems. That makes it the least common denominator in terms of wideband codecs. It uses exactly the same amount of bandwidth as G.711.

There are many other wideband codecs, like; AMR-WB, iSac, G.711.1, G.729.1, G.719, G.721.1, G.721.1C, SILK, Speex & CELT. Many of these have specific applications. For example, AMR-WB was designed specifically for wireless networks (cellular.)Some are patent encumbered and so licensed at a cost. Some are patented but royalty free, so there’s still a cost of entry to get the documentation and use the codec. Some are open source.

Most of these codecs use less bandwidth than a G.711 encoded call. Some much less. Modern coding transforms can be very bandwidth efficient. But low bit-rate has its own headaches. It makes packet loss a more serious issue. Thus there are serious packet loss concealment (PLC) routines that often accompany very tight codecs. These may or may not be free/patented/open sourced.

Historically, open source codecs like SPEEX and the newer CELT don’t see much uptake in hardware devices. There are many reasons for this, most not technological, but legal. No-one wants to invest R&D into a device that cannot be sold because there’s a possible patent challenge against an open source project. It’s not that they’re not fine work by brilliant minds. It’s a risk management issue by corp players.

I’d love to see an open source hardware phone! Anyone wanna write code for a Snom 820? Great hardware, just need some really outstanding firmware to go where the corp minds fear to tred.

ATAs are a problem because they connect to traditional phones. Trad phones are a problem becuase they only dial numbers (see SIP URIs to follow) and usually don’t sound very good.

It’s amazing how much better a G.711 call sounds on a Polycom IP450! It’s the same data stream but a physically and electrically better device. Most analog phones are simply cheaply made, and they sounds like it. But people have not been taught to expect better. In fact, cell phones have taught us to put up with lousy call quality. And fly-by-night VoIP providers reinforce that reality.

Yet, VoIP can and should be BETTER than the PSTN.

VoIP has to get beyond providing x minutes for $y/mo. The cost/minute for calling is racing towards zero. There’s no margin for anyone in that trend. How many companies have to go under to make that point? To provide service takes money to pay people. If margins are shrinking and absolute revenues declining how does anyone provide service? In the long run they don’t.

Wideband voice is better. Period. Don’t believe me, then try it yourself. Use a SIP client to call the ZipDX wideband test at wbdemo@conf.zipdx.com or visit the Gigaset wideband sample web site at »www.gigaset-sound.com.

Oh yeah, to actually use wideband voice you pretty much have to forget about phone numbers. Phone numbers are solely an aspect of the PSTN and likely won’t get beyond G.711.

Imagine if you could only send email to someone if you knew their IP address! That’s the parallel to the PSTN. To go beyond the PSTN we have to leverage peering via SIP URIs, not unlike email addresses.

Remember before email, you had to put a stamp on a letter or pay for a courier or fax. Do you think that the post office or couriers liked the onset of email as the dominant method of communication. I think not.

Phone companies really don’t like any kind of IP peering because it will be the final thing that drives the cost of calling to zero. It will make them irrelevant beyond being one of a number of providers of IP connectivity.

All this seem interesting? Join us on a Voip Users Conference call some time. www.voipusersconference.org

Listen to our call from two weeks ago when Dan Behringer covered the topic pretty soundly.

Or attend the HDComms Summit May 21st in NYC. www.hdcomms.com

Michael