“Sorry…. you’re breaking up… Hello? could you repeat that
please?”
What price Quality of speech? What is acceptable speech quality on a phone
line to you? What influences your
perception at any one given moment?
I often get asked : “why should I pay you for a VoIP service
when I can use Skype?”. I thought I’d
write a bit about my take on VoIP quality.
There’s VoIP and there’s VoIP. There’s good quality voice and there’s not so
good quality voice. When Storm kicked
off our VoIP offering back in February 2004, our Sales team were often faced
with some serious skepticism – especially from IT Managers who had “been there,
done that, wanna see my scars?”, referring to some voice over IP solution they’d
tried and failed. The guys who often
took the plunge in 2004 were the FD’s – keen to cut costs. It did help that our solution was capex free
for the customer and ROI did not need to feature.
Our customers being businesses, many of whom depend on their
voice comms to keep the daily bread coming in – we’ve a fair portion of the BPO
market here – will not tolerate bad quality voice. But the problem is this : quality is in the
ear of the hearer. How often do I witness
people battling to get a good signal on a GSM phone, accepting “poorer than PSTN”
quality, calling back once or twice during a call and seldom thinking of
logging a fault. I do it myself. But let my desk phone be faint, or noisy, or
break-up, and my blood pressure starts to rise! Unconscious perceived value in the mobility of the GSM handset and
acceptance of a norm that has arisen.
In a bandwidth-lean environment like South Africa, where
uncontended last mile IP connectivity for an internet services provider is
still prohibitively expensive (in November 2006), we have had to resort to an
actively managed IP network. MPLS based virtual
private networks with QoS, the order of the day. Quality VoIP, but at a price.
So it was with interest that I read two articles, reporting
research into recent trends in the quality of VoIP over the open internet. Both coming to different conclusions.
One, referenced on the Gartner blog, quoting a recent VoIP
study conducted by Brix Networks
indicating that VoIP (not over a managed IP network) quality is erratic. The
study asserts that quality has gone down over the past couple of years, but
Gartner’s reading of the data leads them to think otherwise. They point out that a cut of the most recent
data shows that VoIP quality is improving – but then it’s still 5% down on Feb
’05. Brix have a cute little app you can
run on your desktop that measures and plots MOS scores to various difference
global destinations – I ran it for a day or two until Google Desktop (which is
it requires) irritated me enough to get rid of it. Some of the screenshots I’ve pasted into a Word doc if you’re interested - check under "Recent Presentations".
The 2nd, a study by Minacom, seems to indicate
that “VoIP phone service worldwide "now sounds better and connects faster
than the standard public-switched phone network (PSTN)."” Better than the PSTN?!? They go on to point out that Brix’s tests
were based on PC-to-PC type of VoIP, rather than VoIP such as we at Storm
deploy, where we manage the IP network from the Gateway attached to a PBX to
the last point before it is converted into a TDM signal again (if at all).
Better than the PSTN though? That brings me to an interactive white paper I love to quote (it’s a bit
like a party trick for geeks! ;). The
guys at Global IP Sound (who developed the iLBC codec used by Skype) put it
together. I love it. VoIP has the potential to deliver broadcast
quality audio on a voice call if all the elements (including bandwidth) are in
place. PSTN cannot and will never. In the interests of squeezing every last drop
of bandwidth out of the “pipes” we use, we purposefully degrade the quality of
a VoIP call to the point where it is efficient, yet only just acceptable. That’s the line we have to walk, until the
price of decent quality bandwidth falls..
That’s me.
Btw. If anyone can point
me to the results of some guys who have been doing regular pings around the
world for the last decade or so and publish the results, please drop me a line.
