I know you asked for interpreting, not CART, but some of the issues are the same, so I thought I'd share.
I've used remote CART a few times. Never in class - my university has a relationship with a local group of CART providers who only do non-remote CART - but I've been to a few events that were remote CART'd.
The set up is basically this: the interpreter or provider is present via two programs: one displays the text they're CARTing (I think via a web-ish interface - some sort of AJAX-y HTML, perhaps?), and the other is a VOIP phone, usually Skype. I've never been involved with setting it up, so I don't know if they provide the laptop, or if they just send the organizers a disc with the software and tell them to set it up on their own laptop. Anyway, they get called on Skype maybe 5-10 minutes ahead of time, and test that they can hear what's happening at the event, and that their text is showing up on the screen.
I'm a big believer in VOIP and Skype; but they're not ready for this yet. Nor, I think, would it be any better using a regular landline phone. Disconnections are common; even more common is that the provider can't hear well enough to provide the sort of service that they would onsite. The incentive to provide this sort of service is obvious; it's much cheaper for consumers, and involves less travel for the provider. But as far as I have seen, it's a much less polished 'product' than can be provided in person.
*If* you had only one lecturer, who was mic'd, and *if* the interpreter/CART provider was thoroughly briefed ahead of time on the topic, I could maybe see it working out. But getting contextual clues from what you can see makes a huge difference, and I don't think the audio tech is ready yet to deal with a lecturer, questions from the audience, and so on and so forth. That said, if I used interpreting rather than CART, I might be willing to try it with a small class - say, no more than 15 people - in an easily-mic'able room configuration, if I had the option of saying later in the semester: "hey, this isn't working. Can I get a live terp?" Still, most of my classes are at least 30 people. (Also: if I was doing this, I'd consider a text chat function a major bonus - some way to feed an interpreter new vocabulary being written on the board.)
Not really what you asked for, I know, but since not many others have answered ... *shrug*