hifi-lofi / audiofft Goto Github PK

virtual ~FFTW3FFT()
{
  init(0);   <= It seems that 'fftw' resources will not be freed in such destructor.
}

Because in 'init' function deallocation will be done only if 'init' parameter is greater than zero.

Hi there,

In your README in your example code, when you initialise your real and imaginary arrays you write:

  std::vector<float> re(fftaudio::AudioFFT::ComplexSize(fftSize)); 
  std::vector<float> im(fftaudio::AudioFFT::ComplexSize(fftSize)); 

but your namespace is called 'audiofft'.

A tiny thing, but it would save others a few mins to fix it, hard to spot! :)

Great library BTW!

Thanks!

Adam

Dear,

As you say,
if I use the nfft=2^16,the time will more than 10s, is it ?

Thx

Hi,

First of all - thanks for making the AudioFFT under the MIT license.

I am trying to add support for Intel IPP.

My issue is:

• I can't find anywhere in the documentation of the IPP, what is the size of the DFT produced by their algorithms.

As much as I can see, it is the same size as the input, which means it preserves the conjugate symmetries above the N/2 + 1 mark. There are functions that would compute the DFT with "packing" to what seems to be N/2 + 1 length, but they don't produce complex results... and this confuses me...

Any chance you have used the IPP?

Cheers!

I had a question...while calculating fft of input samples, does this code takes arbitrary size of input and calculate say 4096-fft, because apparently it calculates only if the no. of input samples is same as the fftsize? Please help..

So that it can be used at compile time to declare an std::array for example

Hi.

Nice job with the library. I assume the fft goes until 2pi?

Thank you.
Marcos