To be honest, not all the facial animation is uniquely recorded - at least the way I understand the tech, there are two stages. First they capture color + normal maps and geometry for every possible kind of facial expression (based on some system of elemental small expressions) and build a facial rig. Then they record audio and facial movements to drive the rig, at the same time.
There's also the separate session for the body mocap, which of course brings up sync issues but those are even more problematic in scripted gesture-based systems like Mass Effect.
Then again I might be wrong at it may really require completely unique capture data for every second of facial animation... but that'd mean a huge amount of data!
The real question here is how much of the gameplay these interrogation scenes are. If there's driving, shooting, puzzle solving / evidence searching, then it should still be more then Dragon's Lair.
And every other RPG today relies on prerecorded acting too, it's just that they only record voice which is probably a lot faster and easier to work with, so they can offer more content. I'd say the Mass Effect games still has like less than 40-60 hours of voice altogether, even if you combine all the possible dialogue options for both the male and female protagonist. Of course there's also the entire combat and exploration part too...