This IR protocol is presumably Kaseikyo, which is used by many Japanese manfacturers.  The first two bytes of these signals are 20.99, which is Fujitsu's Kaseikyo OEM ID.
So the irp should start {36.7k,432}.  36.7KHz is certainly correct, but the 432 uSec was measured from lots of learns of Kaseikyo signals, and so it may not be quite correct.  Additionally, Pronto Hex and perhaps other formats have trouble getting these durations entirely right, so some experimentation is necessary.  It's difficult to know whether to specify the on duraton as the envelope of the on burst (413) or instead to spec the time as a multiple of the modulation period (432). However, changing from 413 to 412 is really too small a change, so the results are likely to be unstable.  Give a longer time a try.