Compose FFT

Hello

FFT can convert a given number of samples from time domain to
frequency domain.
Ok... what about the inverse?

Basically I'd like to have something like that:

input -> FFT -> "something" -> output
with input = output

I think I could do that generating sinusoides on the
frequency/amplitude given by FFT and summing them, but it probably
will be very slow

Any ideas?

thanks

Luis Fernando wrote:
> Hello
> 
> FFT can convert a given number of samples from time domain to
> frequency domain.
> Ok... what about the inverse?
> 
> Basically I'd like to have something like that:
> 
> input -> FFT -> "something" -> output
> with input = output


You are describing an inverse FFT.
	ifft( fft( x  ) ) == x

The mathematics are almost exactly the same as the forward FFT.

I'm curious.  How did you learn about the FFT without being exposed to 
the inverse FFT?


-- Mark

now I found something called IFT that probably do what I asked ;)

trying to work with it...

Well, the FFT is invertible - just use the iFFT. But since in your
application input = output, there is probably no reason to use it at
all...?

--smb


wavebox@wavebox.com.br (Luis Fernando) wrote in message news:<fb78d2e7.0407051418.72ae4004@posting.google.com>...
> Hello
> 
> FFT can convert a given number of samples from time domain to
> frequency domain.
> Ok... what about the inverse?
> 
> Basically I'd like to have something like that:
> 
> input -> FFT -> "something" -> output
> with input = output
> 
> I think I could do that generating sinusoides on the
> frequency/amplitude given by FFT and summing them, but it probably
> will be very slow
> 
> Any ideas?
> 
> thanks

in fact I didn't learn that
i'm still learning many things in a very short time and i was told to
make an FFT... then I found the FFT function on VisualDSP++, called
dit and it worked ;)

Mark Borgerding <mark@borgerding.net> wrote in message news:<UslGc.186055$DG4.167479@fe2.columbus.rr.com>...
> Luis Fernando wrote:
> > Hello
> > 
> > FFT can convert a given number of samples from time domain to
> > frequency domain.
> > Ok... what about the inverse?
> > 
> > Basically I'd like to have something like that:
> > 
> > input -> FFT -> "something" -> output
> > with input = output
> 
> 
> You are describing an inverse FFT.
> 	ifft( fft( x  ) ) == x
> 
> The mathematics are almost exactly the same as the forward FFT.
> 
> I'm curious.  How did you learn about the FFT without being exposed to 
> the inverse FFT?
> 
> 
> -- Mark

yes
of course that's a test ;)

spam@dspdimension.com (Stephan M. Bernsee) wrote in message news:<42c34ef.0407052023.24174d2a@posting.google.com>...
> Well, the FFT is invertible - just use the iFFT. But since in your
> application input = output, there is probably no reason to use it at
> all...?
> 
> --smb