We are testing the newly baked SSE modifications FFT / MDCT

I wrote for my creation SSE-implementation of complex FFT and MDCT. The results even amazed me:

conversion / option	without SSE	SSE modification
2000000 * FFT-128	16.7s 13.3s	5.1 s 3.7 s
1,000,000 * MDCT-512	17.4s 15.3s	10.4 s

Two million 128-point complex FFTs and one million 512-point MDCTs were calculated. Intel® Core (TM) 2 Duo CPU E8400 @ 3.00GHz

I wonder if this can be sold or licensed to someone? :)