Sound Bytes From the Field
Users are not experiencing HD voice; they are definitely experiencing the PSTN and mobile networks. Call quality complaints are nearly non-existent--thus far.
HD Voice seems to be silent these days, but for those that have interconnected systems or platforms supporting HD, my guess is that users do notice. Still, I wonder what is their perception of voice quality and whether or not it is acceptable to have as-good-as or almost-as-good-as PSTN services? Then, when customers experience HD, do they notice the difference when they subsequently switch to the PSTN?
In the past, I've noted that we aligned our two providers, Broadvox and Skype Connect, to use the same compression (Codec) choices--G711 first choice and then second choice G729. Before I address call quality, I want to mention call completion. Every call offered to Broadvox has successfully completed, and I can't recall any instances or complaints of incomplete calls or fast busies. With Skype, we've noted that certain exchanges with mobile phones and numerous toll free numbers pegged a high amount of incomplete calls and users receiving a fast busy or reorder tone from Skype Connect.
We've tracked both trunk groups, Broadvox and Skype, and pulled the average MOS scores for all calls in each group. Here's what we found:
MosLq: is the listening quality MOS. Listening quality indicates the perceived quality of the transmission for a user not actively involved in the conversation, but passively listening. Listening quality does not consider delay or recency.
MosPq: is the listening quality MOS normalized to the PESQ scale. Users interested in scoring based on the ITU-T recommendation P.862 (PESQ) should use this value.
In another project in a different environment we recently worked around the clock installing a temporary solution to minimize impact on a customer. A mesh network using tunnels to each site was created using Comcast as the provider with 50Mbps/10Mbps service at each location. Each location is situated within the same MSA (Metropolitan Statistical Area). Here's what we saw:
What did I learn from these statistics? First, MPLS may be the more expensive service but it doesn't mean that "acceptable" call quality isn't possible with broadband providers. Granted, the three sites above are geographically within the same MSA, and MPLS does assure QoS, but what about these tunnels? With no user complaints and inter-office calling and faxing continuing to flow, what does it indicate?
Users in my company and customer sites are not experiencing HD voice, and maybe that's a good thing for me. On both sides they are definitely experiencing the PSTN and mobile networks. Call quality complaints are nearly non-existent--thus far. (I'm knocking on my wooden desk as I type.) Do the lower MOS scores indicate that we can't get HD voice? Other than some calls rejected by Skype, audio quality just isn't a discernable or notable difference. Using Comcast to create a mesh network proves at least call quality is supported and achievable within geographic confines. What really gets my attention is the difference in costs. The Comcast links are about $135 a month.
In our VQM (Voice Quality Monitoring) what I did do was set monitoring for all RTP flows on both the LAN interface and the tunnels. Why? Curiosity! When the calls leave one physical interface (Ethernet to/from Comcast modem) and go into that virtual world (tunnel), is it safe to assume that call quality remains constant?