High-Definition Voice: The Future of Phone

Texas Instruments envisions a world where voice is a part of every IP connection—and drives our lives.

By Ted Stevenson | Posted May 23, 2007
Print ArticleEmail Article
  • Share on Facebook
  • Share on Twitter
  • Share on LinkedIn

Isn't it odd that, nowadays, we expect DVD quality in the videos we watch, CD quality in the music we listen to, and, increasingly, HD quality in the TV we watch, but when it comes to phone calls, we settle for the sound produced by a technology that's almost one hundred years old?

It's true, though. Even though parts of it are now digital, our public switched telephone network remains constrained by a set of standard assumptions and technologies that really haven't changed for decades.

Anyone who's bought a decent headset and signed onto Skype or GoogleTalk or Gizmo Project knows that IP telephony is capable of fairly breathtaking sound improvements compared to traditional PSTN quality.

The phone folks at Texas Instruments know this too, and they're bending their efforts to changing the sound landscape of telephony—and changing its role in our lives as well.

VoIPplanet.com recently spoke with TI's IP phone product manage, Angela Raucher, who filled us in on the company's efforts to create and promote high-definition audio technology.

What is "high-definition voice"? There are two ways to answer that question: by analogy, and with technical specs. According to Raucher, "the difference in the audio quality that the user will experience is like the transition from AM radio to CD-quality stereo."

On the technical side, traditional telephony is based on sampling the sound stream 8,000 times a second, and constraining the reproduction of the sound spectrum to the range between 200Hz on the low end to 3.3KHz on the high end—and fitting it into a 64Kbps bandwidth. A "wideband" codec (coder/decoder algorithm) doubles the sampling rate and more than doubles the width of the sound spectrum reproduced, from 50Hz to 7KHz. This adds significant depth and nuance to the transmitted sound—and it reduces the bandwidth requirement to 32Kbps, half that of PSTN transmission.

Apparently TI is not content to stop there, however. "We just made an announcement that we're licensing a super-wideband codec," Raucher explained. The codec in question—the MPEG-4 AAC Low Delay codec, developed by the Erlangen, Germany-based Fraunhofer Institute for Integrated Circuits IIS—boosts the sampling rate to 48k per second, for near-CD quality at data rates of 48–64Kbps.

TI will build the algorithm into its line of high-performance digital signal processors (DSPs), another key ingredient in the recipe for high-def voice. "We're preparing our software framework for rich multimedia," Raucher said.

So, the combination of advanced super-wideband codecs and huge processing power (systems on a chip capable of processing 100 sound streams simultaneously)—together with the ever-growing data capacity of IP networks, even in the home—will significantly change users' expectations, leading, as TI sees it, to a world in which virtually every IP connection will be voice-enabled.

High-definition VoIP will find its way into a myriad of devices. "We're working with the service providers who are looking at bringing out DSL modems with voice, cable modems with voice, standalone gateways, fixed/mobile convergence products," Raucher said.

How long will we have to wait for these devices—and applications? The issue, according to Raucher is at least partly one of cost. "This kind of stuff's available today," she said. TI already produces a "system on a chip, with the capability to do wideband conferencing," video, and other nifty multimedia apps. But it's pricey. Three to five years down the line, she predicted, you'll be able to buy phones and related devices built with the high-end components—only more highly integrated than any available today—for less than the current cost of a standard desk phone.

Where is all this leading? TI is looking forward to the day—at some unspecified point in the future—when, as we mentioned earlier, every IP connection will not only be voice-enabled, but voice-enabled with true HD capability.

The enhanced sampling and fidelity of HD codecs will lead to improved accuracy of voice-recognition applications, which, in turn, will lead to a "universal voice interface" for our computers, appliances, and other systems—piped over those ubiquitous IP network connections.

Comment and Contribute
(Maximum characters: 1200). You have
characters left.
Get the Latest Scoop with Enterprise Networking Planet Newsletter