Thanks for your input. Yes, we’ll bear delay in mind. It seems to me that the whole problem with researching VoIP is that almost all published works talk about it at a conceptual level. There is nothing which really goes into the design requirements in sufficient detail for us techies 🙂 – an imbalance we would like to correct with our new site.
Perhaps this is because the technology itself isn’t mature enough. Using your example, it is difficult to quantify the delay effects of an end to end transmission, because the effects of quality of service initiatives seem so vague. However, the delay characteristics of various CODECs are documented, and perhaps that is good starting point.
Voice over ATM is not in our immediate plans, but is something that we will keep within our sights.