it's ivr + landline only.
the last canadian election had a company doing polling with a similar method (ivr) and with the same argument that it "reduces bias" (which is just a guess, really, and not a good one). they consistently had the conservatives five or six points higher than anybody else. and, they ended the election with the worst results.
i don't think that ivr "reduces human bias". rather, i think that what it does is it reduces the sample space and creates a distorted outcome.
their results are just widely deviant. and, i'd be careful with them.
the best polls are with human interviewers. this is very well understood.