I'm using a good fast FFT implementation (vDSP) that will only work on power of 2 blocks of audio data. Now I have a problem where I would like to be able to apply the calculations to non powers of 2 audio.
Some degree of reading around suggests that I can do this by breaking the DFT into smaller power of 2 blocks.
Can anyone explain how to do this, if indeed it is possible at all? I have been unable to find any explanations of how to do it ...