a couple of off-the-wall notes. i'm bouncing off of stephen sprunk here, but in fact these notes have very little to do with his excellent article:

Toll-quality voice requires ...

...all kinds of things that nobody outside the POTS empire actually cares about. folks just want to talk. cell-quality voice is fine. (just ask anybody in panama who has relatives in the USA!) sadly, to get "voice over ip" (note, it's not telephony over ip, it's voice over ip), we're going to have to integrate it into our computers. ("dammit, i need a decent quality USB headset for less than USD $300!") because as long as something looks-like-a-phone, the POTS empire can use the NANP (or local equivilent) and 911 regulations (or local equivilent) to prevent newer more efficient carriers from making money from "voice". the solution of course is to use computers rather than "phones" and to use domain names rather than "phone numbers". "old money" has a way of flattening the decline of old technology... until the cliff, that is.

..., the public Internet has substantial jitter and high coast-to-coast latency, ...

just thinking out loud here, but which "coasts" do we mean when we talk about the "public internet"? my first thought was lisbon-to-sakhalin, rather than seattle-to-miami. given that the public internet isn't even centered in let alone predominated by north america any more, and that some of the best (and/or loudest) speakers at nanog (both on the mailing list and in person) are from outside north america, it seems to me that the "reform party" should be thinking of a new name. i'll happily turn ANOG.$CNO and/or WORLDNOG.$CNO over to any elected board who becomes merit's successor-in-interest over "nanog governance"... (i'm one of those who loves merit in their secretariat role but who thinks that issues of list/conference content/moderation, budgets, program committee membership, and overall policy should be handled representationally, like at ARIN.) (if you didn't know about the nanog-futures@ mailing list, go find out, plz.)

OTOH, if you're going across a network with decent QoS or within the same general area of the country, you can afford a larger transmit buffer without risking the "walkie talkie" effect.

all it has to be is as good as a cell phone. that's a lot more than 125ms.

...

I mean, where would you set it, and can you offset some of that with fragmentation and intervleaving?

F&I is a technique for reducing jitter on slow, congested links like the last mile to a customer. It's often combined with a priority queue, since the latter is not enough on such links (but is on faster ones). Neither has much to do with the (tiny) sizes of voice packets.

somebody told me that they can get about 8 concurrent G.711 calls over an 11Mbit/sec 802.11 link, and maybe 9 if they use G.729. apparently the radio link arbitration on 802.11 is better than ALOHA but not as good as "thickwire", and the packets-per-second maximum is low-in-practice, and we'll only see full rate (11Mbits/sec, 54Mbits/sec, whatever) from single-flow-large-MTU. that's ok for me, it's still good enough for a horsecam and a barnphone. but it's something worth checking into for anyone who plans larger deployment.