[Octal-dev] Fourier Transforms and samples

[Steve Mosher]

        I've been thinking about the pitch-scaling / time-dilation issues,
and learning some cool stuff about math at the same time. Performing
Short-Time Fourier Tranforms on each sample played causes performance
problems, but is the bulk of it in the transform itself? Or is reassembly
similarly intense? I'm going to look at the reassembly process more
closely, so I can see this for myself. Anyhow, what I'm getting at is
this: would it be worth it to store selected samples as transformed? Less
work would have to be done in real-time... in fact, all STFT compliant
machines could neglect to reassemble their data between one-another.
Regardless, the signal must only be transformed once, and reassembled
once, which is good in situation where you want to (say) scale the pitch,
then put it through a freq-filter. 
        Does this work out well, or am I missing something?

-- 
        It is necessary for the welfare of society that genius should be
privileged to utter sedition, to blaspheme, to outrage good taste, to
corrupt the youthful mind, and generally to scandalize one's uncles.
        -- George Bernard Shaw