From: Jim Peters (jim_at_uazu.net)
Date: 2002-02-27 10:36:07
Joerg Hansmann wrote:
> Always. I did not find an option to avoid the interpolation or
> change the type of interpolation.
>
> See sampleindex 10559 to 10560 for an example. (attached picture)
The 'mathematically correct' process for upsampling to a higher
sampling frequency is actually to turn the samples into impulses, and
then put a perfect step low-pass filter over them at half the original
sampling frequency. If I remember correctly, this is equivalent to
convolving with the sinc ((sin x) / x) function. This can give
overshoots as you're seeing.
If this is what they're doing, you could actually be seeing the
'mathematically correct' interpretation of the samples. Other forms
of interpolation (linear, or fitted-curves, say) always give
distortion of some kind.
(All of this according to my present level of DSP knowledge, so query
me if this doesn't make sense)
Jim
-- Jim Peters (_)/=\~/_(_) jim_at_uazu.net (_) /=\ ~/_ (_) Uazú (_) /=\ ~/_ (_) http:// B'ham, UK (_) ____ /=\ ____ ~/_ ____ (_) uazu.net
This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:39 BST