11 Dec
2002
11 Dec
'02
4:51 a.m.
I'm trying to oversample to smooth the display on my software audio
scope using band limited (sinc) interpolation. I have a quick question
to ask. Which of the following implementations is liable to take more
processing time:
1) Padding my data, and then convolving the original (non-zero) samples
with a sinc function (from a LUT). I think it's only feasible to have a
LUT for a sinc of unity amplitude.
or
2) Doing an FFT on my original sample set. Padding, and multiplying by a
rectangular window, then doing an iFFT to return to the time domain. The
FFT library I would use would be FFTW.
Cheers
Henry