Communication breakdown threatens VoIP

TalkBack 10 of 11:

Next »

« Previous

• Thread View
• Flat View

Bu that;s not all..

An excerpt from the story..

"As a result softswitches and Internet protocol gateways from one vendor may not interoperate with equipment from another"

Granted, this is one problem but I'm in both the telephony and speech recognition marketplaces and would point to other "network-to-network" issues which seem to be largely ignored and the (opinion) VoIP industry will suffer stunted growth as users convert with often dissapointing results, until they're cured.

Nyquists's sampling theorem states:

"If a signal f(t) is sampled at regular intervals of time and at a rate higher than twice the highest signal frequency, then the samples contain all the information of the original signal"

The telephone industry which largely defined the standards carrying over into VoIP ignored in it's infancy, for various reasons the bandwidth requirements for "live" voice quality, to wit about 200Hz to about 7 kHz - 8kHz.

Instead, the baby Bells elected to trim down the frequency response of both telephones and the PSTN form approx. 450Hz to approx. 3kHz, under the (then contemporary) thinking that enough significant voice data resided within this bandwidth limitation to maintain "good enough" voice quality and it was of course much cheaper, as well.

This frequency response *was, for the most part "good enough" back when we were transmittiting /receiving pure analog across twisted pair & hardwired networks in the good old days.

However, when carriers began converting PSTN infrastructures to digital networks, again the baby Bells elected for economic (and then contemporary thinking) reasons to go with a less expensive, easier to implement sampling rate for digitising voice to reproduce "telephone-quality" conversation - 8KHz; twice the 4KHz required for what was then believed to be the full spectrum of the human voice. However, in the real world this 8kHz sampling does not capture the even 3kHz bandwidth; the 8-th bit (least significant bit) of the sampled word is discarded and this bit position is used to carry signaling and control information for the system.

As a result, the digitizing does not render the entire 3kz; in all actuality, almost two bits are lost in the ADC process due to the above and additional factors, commonly referred to as "overhead".

Ergo, the "standard" digitizing of our voice renders closer to a 6 bit word, only further crippling the voice reproduction quality of what's being transmitted across networks even today.

We now know, and have for years that to properly capture all the sibilants, et al of the human voice a much larger bandwidth is required (7-8kHz), and an accordant increase in the sampling rate to at least 16kHz (Nyquist, above) is necessary as well.

As the sampling rate is increased, so too is the quality of the digital conversion. The faster the sampling rate and the larger the sample size, the more accurately the sound can be digitised.
Because the technology in most VoIP networks still remains rooted in old thinking (3kHz/8 bits) hence the often "tinny" and less than satisfactory voice reproduction heard across VoIP networks. When using the Mean Opinion Score (MOS), created by the ITU most voice reproduction across networks just doesn't make the grade.

The other intrinsic problems that plague the interoperability of VoIP networks - most Internet protocol gateways are built around this outdated frequeny response/sampling and even when the originating transmital instrument (phone/cell phone/PC) are rendering a far better packet, 16 or preferrably 22kHz/16 bit signal, gateways tend to ignore what's presented and will re-format the signal to the familiar (and most common) 3kHz/8 bits packets, nonetheless.

Yet another problem is data compression necessary for networks to handle the traffic - although PCM and later some ADPCM is used to compress the digitized voice, there exists no real standard yet, and there should be: various VoIP vendors use different compression, alogorithims, echo cancellers, jitter control, silence suppression etc.. Although excellent voice quality is possible at rates as low as 4.8kbps, most networks just don't seem attracted to the development of transmission technology and algorithms that will deliver the right balance to meet bandwidth and voice quality requirements.

Posted by: wmburke   Posted on: 01/15/04 You are currently: a Guest | Members login | Terms of Use

Alert moderator to an offensive message

Subscribe to this discussion via Email or RSS

SponsoredWhite Papers, Webcasts, and Downloads

• Three Steps You Need to Know to Stop Data Loss Varonis Sensitive data exposed to misuse or loss... it is the stuff of nightmares ... Download Now
• Building the Virtualized Enterprise with VMware Iinfrastructure VMware VMware virtualization software has been adopted by over 120,000 enterprise ... Download Now
• The Impact of Virtualization Software on Operating Environments VMware Today's use of virtualization technology allows IT professionals to ... Download Now