The FDR showed the throttles advancing to full power a bit less than a minute before impact. At the (roughly) 5 degree "dive" or "approach" angle I calcultated in another post (with John somebody) from the plot (2000 feet descent to impact in roughly 4 nm from the end of the 360 degree turn) and in a clean configuration, I don't know if a 757 would accelerate to 430 knot or 460 knots in a minute (I'd have to ask a 757 pilot) but it would still be accelerating at impact. I don't have a fine enough resolution on the graph from the NTSB release to do a modest time distance guess. I also don't know if during the original descent from 7500 feet the pilot wasn't accelerating during the turn, or descending at constant airspeed (which is the normal way to descend and airliner when you aren't trying to use it as a missile.)
The initial speed estimates may have been based on a gear down configuration, which considerably increases drag and reduces max speed.
One could estimate pretty closely based on film footage that was time tagged in tenths of seconds, and expert displacement measuring, but I don't know if that was done, or if such film exists.
At impact, the critical dimension is groundspeed, speed relative to the ground/fixed building's frame of reference, not airspeed.
DR