[MPlayer-dev-eng] [PATCH] fix for -srate bug

[Svante Signell]

On Wed, 2004-10-27 at 23:00 -0400, D Richard Felker III wrote:
> On Thu, Oct 28, 2004 at 12:38:07AM +0100, Ed Wildgoose wrote:
> > 
> > >btw, iam very interrested in any feedback about its quality especially if 
> > >it includes a hint on how to improve it
> > > 
> > >
> > 
> > I have been flicking through your code a little.  Do you have any 
> > references to the algorithm you are using?
> > 
> > From my limited understanding of the code, I can't really see how you 
> > are getting the claimed high quality results?  For example on 
> > upsampling, are you actually bandlimiting?  If not, then aren't you 
> > going to get horrendous aliasing?
> 
> how the hell do you get aliasing when upsampling? aliasing comes from
> downsampling without bandlimiting.
> 
> rich

Sorry for jumping in here but I have a few comments:

When doing upsampling the unwanted bands are called imaging bands. They
have a similar effect as aliasing when downsampling. Unfortunately in
both cases you have to filter _after_ the upsampler and _before_ the
downampler, i.e at the high speed side. However, efficient realizations
can be made by using a poly-phase structure of the aliasing/imaging
filters. Maybe this is of small value here, since we are talking about
conversions 44.1kHz to/from 48kHz, or?