This should generate fft blocks from the audio data.

You need a copy of the FFTW library to build it.

Output data is an ntsc and pal data block, fft is in 1024 * unsigned short
blocks. The imaginary and negative components have been removed so...