Hi everybody,

I'm using ADSP-21375 and I'm trying to make some manipulations on audio data, but I have some problems.

I want to apply fft on the first input, make some maths and then apply ifft to back in time domain. At last, I want to make a convolution with the second input. I'm not interested in real-time.

Problem using filter.h:

- I can't use convolve function with complex_float type.

- Can I write out_complex.re and out_complex.im to get the real and imaginary part of the output of the rfftN?

Problem using trans.h:

- This header does not contain a convolve function.

- Moreover, I read that rfftN and ifftN functions take the input data reversing the bit. Does it mean I have to flip the input and/or the output? This point makes me confused.

Thank you very much for your help!

Andrea

Hi Andrea,

Please see my reply below:

Problem using filter.h:

- I can't use convolve function with complex_float type.

>> Unfortunately, we don't seem to have a function to convolve complex_float type of data.

- Can I write out_complex.re and out_complex.im to get the real and imaginary part of the output of the rfftN?

>> Yes, out_complex.re and out_complex.im will provide the real and imaginary part of the output.

Problem using trans.h:

- This header does not contain a convolve function.

- Moreover, I read that rfftN and ifftN functions take the input data reversing the bit. Does it mean I have to flip the input and/or the output? This point makes me confused.

>> Are you referring to the file "fft.asm" under the pah "C:\Program Files\Analog Devices\VisualDSP 5.0\214xx\lib\src\libdsp_src" ? I think idoes the bit reversing in the begining as it implements decimation in time FFT algorithm. But, it may be transparent to the application.

Hope this helps.

Thanks,

Mitesh