Pillars of VoIP service - bandwidth, architecture

For all of the high-profile "Broadband VoIP" advertising we're being inundated with of late, you'd think that the major differentiators between leading contenders fall into just two categories: bells & whistles, and price. But what about voice quality? Does it matter anymore?

A quick review of the home pages of high-profile VoIP services Lingo (Primus), VoiceWing (Verizon) and Vonage (at this writing) throw more features and pricing information than you can possibly absorb but not one of the three ever makes reference to voice quality.

Why might that be? Can they consider the concept so esoteric that it is above the average consumer at whom their sites are targeted? Hardly. Then perhaps they don't mention it because they've determined that it is not a factor in service selection - that it simply is not a differentiator.

To continue reading, register here and become an Insider. You'll get free access to premium content from CIO, Computerworld, CSO, InfoWorld, and Network World. See more Insider content or sign in.

For all of the high-profile "Broadband VoIP" advertising we're being inundated with of late, you'd think that the major differentiators between leading contenders fall into just two categories: bells & whistles, and price. But what about voice quality? Does it matter anymore?

A quick review of the home pages of high-profile VoIP services Lingo (Primus), VoiceWing (Verizon) and Vonage (at this writing) throw more features and pricing information than you can possibly absorb but not one of the three ever makes reference to voice quality.

Why might that be? Can they consider the concept so esoteric that it is above the average consumer at whom their sites are targeted? Hardly. Then perhaps they don't mention it because they've determined that it is not a factor in service selection - that it simply is not a differentiator.

While some would say that cell phones have lowered our expectations when it comes to voice quality (and I wouldn't disagree), I don't think anyone would say it doesn't matter. In the case of cell phones - and despite claims of great coverage implied by the "can you hear me now?" ads - it's virtually axiomatic that every user will experience some poor-quality voice on some occasions. And, with so many cell-phone-user-to-cell-phone-user conversations, who can say which service is really to blame?

So true or false?: The reason broadband VoIP services providers spend so little effort on marketing voice quality is because they are all the same.

False. Here at The Tolly Group, we've spent some months setting up and benchmarking the voice quality of a dozen or so such VoIP services - the services mentioned above plus others such as 8x8, AOL, Broadvoice, CallEverywhere (Belkin), Net2Phone, SIPphone, Skype and Yahoo.

If there's one thing that we've determined, it's that voice quality matters. In our research, we used the sophisticated Voice Quality Tester (VQT) from Agilent to measure voice quality using the PESQ scale. Within each service - tested using various samples multiple times at various times of the day - the results were consistent.

But the results varied dramatically across the services tested.

Fortunately, most of the services tested delivered voice quality that rated a "good" or better using PESQ. A few achieved "excellent" ratings - but a few had top scores that were in the "poor" range.

Along with our voice-quality tests, we captured the packets constituting the VoIP conversation and analyzed both the bandwidth consumption and the conversation architecture. As much as anything else, these appear to be the keys to quality.

Voice quality was directly correlated with bandwidth consumption. Translation: If you are willing to devote more bandwidth to your conversation, you'll get better quality. Vonage is one of the few companies we saw that explicitly acknowledges this trade-off and lets its users decide (although not on a call-by-call basis). Users can log on to their account and ratchet the quality/bandwidth up or down as they please. Sure enough, our tests show that when you squeeze the conversation over a narrower pipe, you pay the price in quality.