Should You Heed
the Call for HD Voice?”>
High-definition voice over IP—as exemplified by Polycoms HD Voice technology—provides outstanding quality and clarity for calls within a companys borders and could be an important building block for a range of applications in the future. However, shortcomings of the Public Switched Telephone Network—and of the telecommunications industry in general—limit the current usefulness of the technology outside the corporate network.
This became clear to me as I stood in Polycoms isolated demonstration booth at the Spring VON show in San Jose, Calif. I was immediately struck by the quality of the sounds produced by Polycoms SoundPoint IP 550 and IP 650 HD Voice phones. When compared with a toll-quality implementation side by side, the HD Voice transmissions were perceptibly richer, fuller and clearer.
A combination of technology and education helps address VOIP threats. Read more here.
High-definition VOIP gets its full sound in a few ways.
First of all, high-definition VOIP uses wideband codecs between endpoints.
Normal telecommunication samples a range of audio frequencies (from 300Hz to 3,400Hz) for transmission across the limited bandwidth afforded in the PSTN. Early-generation IP phones followed suit, as the codecs most frequently used with VOIP (such as g.711) were designed to meet but not beat the expectations for toll-quality voice. Polycoms HD Voice phones, on the other hand, can sample sound between 150Hz and 7,000Hz, and the sound is then transmitted via wideband codecs (in this case, g.722), which can support this wider range.
Since the human voice starts at a base frequency of about 100Hz and extends up toward 8,000Hz, the improved sampling of the high-definition technology has a number of potential benefits.
For example, users will expend less energy deciphering sounds—particularly easily confused consonants, such as the letters “f” and “s.” This will lead to better overall comprehension and less fatigue for callers who spend hours a day on the phone. Deciphering foreign accents or dialects should take less effort as well, fostering improved international dialogue.
Beyond the compression and encoding algorithms, high-definition VOIP requires some advanced engineering on the physical phone. Improved audio components and enhanced echo cancellation or noise suppression provide better sound quality (no matter what codec is used), but these features can also help avoid the rumble in the low frequencies or acoustic feedback at the higher frequencies that wideband codecs will open up.
As impressed as I was with Polycoms HD Voice sound quality at the time I saw it demonstrated and during eWeek Labs tests, I was struck by the overwhelming limitations of the technology.
The reliance on wideband codecs means that HD Voice is a boutique affair. The endpoint devices must support the proper codecs for full benefit, but Polycom offers HD Voice only in its pricier and fully featured models. Lower-end models designed for cubicle workers are on Polycoms HD Voice road map.
More disconcerting is high-definition VOIPs lack of extensibility in general: Under no circumstances will high-definition VOIPs full audio quality be maintained if a call touches the PSTN. Any time a call gets routed through the PSTN, whether through a corporate gateway or when peering between service providers, the call will immediately be down-sampled to toll quality, thereby losing all the additional information carried by the wideband stream.
While SIP (Session Initiation Protocol) trunking provides some promise of relief for this problem, high-definition VOIP support will most likely extend to the carriers own customers because U.S.-based service providers typically fall back to PSTN connections—rather than to IP—for peering between carriers.
HD VOIP requires new measurement criteria. Click here to read more.
Polycom officials admit that their greatest traction with HD Voice comes from the largest companies—those with more than 10,000 seats and often with international offices. These companies will reap the greatest benefits from HD Voice calls placed within their own network and between branches (or countries), relying on corporate connectivity to place calls while avoiding the PSTN.
Companies PBXes will need to support the codecs needed for HD Voice. Polycom officials informed me that HD Voice has been shown to work with the open-source Asterisk project, BroadSoft, Cilantro and Interactive Intelligence systems. However, as Ive seen in tests, mileage will vary.
I was able to enable HD Voice with Asterisk with a few simple modifications, but, as is the case with the main fork of the Asterisk 1.4 distribution, such a step hindered the ability for HD phones to connect with legacy non-HD devices. As I later learned from Kevin Fleming, director of software development at Digium, in Huntsville, Ala., “Asterisks code negotiations currently treat the call legs independently, and thus never renegotiate the initial call leg based on the requirements of the secondary call legs.” However, Fleming continued, “In the Asterisk SVN trunk, we have a G.722 code module, so this problem would not occur, and well be putting that module into Asterisk Business Edition as well.”
The potential for high-definition VOIP technology is certainly intriguing. Phone systems are generally a very long-term investment, and the promise of improvements in the IP telecommunications space within the next few years makes the technology worthy of consideration—particularly as well-designed devices will enhance standard narrowband communications in the meantime.
Digital-signal processor manufacturer Texas Instruments sees high-definition VOIP as a possible foundation for powerful applications down the road.
For starters, the improved voice clarity of high-definition VOIP will lead to more accurate speech recognition.
“If I have high-resolution sampling, I have high-quality audio, and I have speech recognition built into a system or product or service,” said TI Director of Technical Strategy Tom Flanagan, in Germantown, Md. “The ultimate manifestation will be when we have enough bandwidth and high definition is widely used—I could be calling someone in France and having real-time translation going on.”
Such real-time translation is still a ways off, however. Processing capabilities need to continue to improve to a point where the translations can do phrases, rather than individual words, to maintain contextual meaning but without taking so long as to introduce too much delay into the call stream.
What is VOIPs missing link? Find out here.
Of more immediate interest to corporations may be the potential improvements high-definition voice could bring to fixed-mobile convergence solutions that bridge the use of cell phones to the corporate Wi-Fi network. High-definition voice could use a number of wideband codecs other than G.722, including one known as G.722.2, or AMR-WB (Adaptive Multi-Rate Wideband).
Wideband AMR has already been approved for use with UMTS (Universal Mobile Telecommunications System)-based third-generation cellular transmissions. This codec supports wideband audio samples but has the flexibility to scale back with limited connectivity. With support for this codec on the internal corporate PBX and desk phones, the potential exists for rich, high-definition voice calls between mobile workers and those back in the office.
I brought Polycoms HD voice into the lab to see exactly what it would take to get HD Voice working with our Asterisk IP PBX implementation.
Starting with Version 1.4, Asterisk supported the G.722 codec in passthrough mode only. This means that Asterisk can set up a G.722-enabled call between two endpoints that support the codec and then get out of the way, but the server cannot transcode the streams between different codecs for devices with mismatched support. And while Polycom officials are investigating adding support for other wideband codecs, G.722 is the only one supported at this time.
To enable G.722 in Asterisk (our server is based on Version 1.4.9), I simply needed to add a single line to the sip.conf configuration file (allow=g722). I then had to configure each Polycom phone with G.722 as the codec with first priority. With these changes in place, I could make calls between my SoundPoint IP 550 and 650 devices and experience all the audio quality I expected from HD Voice.
However, interoperability with legacy devices was another story. While I could place calls from a non-HD Polycom SoundStation conference phone to an HD Voice-enabled phone using G.711, I could not complete a call in the reverse direction. The Asterisk server would show an error indicating congestion on the server, and the caller participating in the testing would experience a fast-busy signal.
Click here to read about VOIP offerings from smaller vendors.
It turns out that Asterisk 1.4 cannot handle the codec negotiation necessary to complete the call between an initiating caller with priority for G.722 and a receiving caller with priority for another codec. “Asterisks codec negotiations currently treat the call legs independently, and thus never renegotiate the initial call leg based on the requirements of the secondary call legs,” said Digiums Fleming.
Asterisk users can look to the bleeding edge for a resolution. “In the Asterisk SVN trunk, we have a G.722 codec module, so this problem would not occur, and well be putting that module into ABE [Asterisk Business Edition] as well,” said Fleming. “We may also put it into future s800i [Asterisk Appliance] builds.”
In my communications with Polycom officials about this issue, I learned that the company has achieved expected codec negotiations when using an SVN trunk of Asterisk. While this feature will likely not be part of the forthcoming Asterisk Version 1.4, users should be able to look forward to full support in Version 1.6 down the road.
Check out eWEEK.coms for the latest news, views and analysis on voice over IP and telephony.