Before or after all corresponding languages have been introduced? Since there are different teams which handle this (it's never the same team), it's probably more effort and resources than we tend to think.
You don't need the narrative team to do it. Assuming what you said is correct and most of the time the differences are alibiing, which I doubt, you could hire temps to do nothing but listen to the audio and change the text to match it. Its not skilled work either so you wouldn't be paying them a lot anyway.