Light sensor for precise light On/Off detection in time, which hardware?

Hello,

I've no experience yet with sensors and harware interfacing, so I'm looking for advices.

I'd like to make a device that would measure sync of image and sound when displayd on screen (video and film).

I know such devices are available commercially, but I thought it could be a fun project for me to train with Max/MSP and sensors, and could be useful for me afterwards.

So the basic thing is to project a black video, with one white image, and a bip at the same time.
Getting the bip thru a mike and a sound card is easy, and with some care, taking into account driver buffer size and such, a precise timing of the event can be achieved.

For the image part, it might be more tricky:

I only wish to detect a change in light intensity (depending on conditions, the black might not be so dark, and the white, not so luminous, but I guess it can be taken into account).
Also a simple tube could help achieve directionality for the light sensor.

The problem I foresee is more the temporal resolution of a light capturing device:
The flash of light is quite short (typically 1/96 of seconds in a movie theater, a 24 fps white frame being in fact projected twice between two dark period for flicker free vision), so it's roughly a 10ms white, 10ms black, 10ms white, 10ms black.
I'd like to be able to detect such an onset of light at least with a 1 ms precision, so that mean that the light sensor should have a ramp time respecting this. The goal is being able to say: the sound was too late by that many ms, with a +/- 1 ms precision at least.

So what kind of light sensor would fit this specification?

What kind of delay would be then introduced by the interfacing with the computer of such a sensor, would it be easy to know, or measure?

What kind of interfacing hardware to use for this? (Arduino or else?)

Could some hardware handle both the sound and light detection? (bypassing the need of an audio interface).

Thanks for any suggestions.