Actually, if the astronaut started falling from a great distance, the astronaut will see light red-shifted, i.e,, the distant outside universe will appear to run slow for the astronaut.
Suppose that observer A hovers at a great distance from a black hole, and that observer B hovers very close to the event horizon. The light that B receives from A is tremendously blueshifted. Now suppose that observer C falls freely from a great distance. C whizzes by B with great speed, and, just past B, light sent from B to C is tremendously Doppler redshifted. What about light from A to C? The gravitation blueshift from A to B is less that the Doppler redshift from B to C. As C crosses the event horizon, C sees light from distant stars redshifted, not blueshifted.
If observer A, who hovers at great distance from the black hole, radially emits light of wavelength [itex]\lambda[/itex], then observer C, who falls from rest freely and radially from A, receives light that has wavelength
[tex]\lambda' = \lambda \left( 1+\sqrt{\frac{2M}{R}}\right).[/tex]
The event horizon is at [itex]R = 2M[/itex], and the formula is valid for all [itex]R[/itex], i.e., for [itex]0 < R < \infty[/itex]. In particular, it is valid outside, at, and inside the event horizon.
See posts 5 and 7 in
https://www.physicsforums.com/showthread.php?p=861282#post861282
I have since done the calculations using Painleve-Gullstrand coordinates that are vaild even on the event horizon.