On Thu, Sep 4, 2008 at 12:28, Carl-Erik Kopseng <carlerik@gmail.com> wrote:

>> Regarding the downsampling I would like to know if I would get any
>> funny artifacts when downsampling 96kHz material to 44.1kHz (not even
>> division). Would I be better of to convert to 48kHz for 96kHz
>> material?
>
> FWIW, I would think 48 kHz would be a better approach, as you'd be preserving
> (marginally) better quality from the original 96 kHz source (not to mention
> having to mess around with padding bits and other hackery that MPEG uses to
> make 44.1 work).

I read quite a few places (like hydrogenaudio.com) that you generally
get better encodings (less artifacts) by resampling to 44.1 instead of
48khz *when using lame*, because it is optimized for 16bit 44.1khz
encoding of mp3s.

Is libsnd capable of resampling and adjusting the bitwidth from
96khz/24 to 44.1khz/16, or would I, as you said, have "mess around
with padding bits and other hackery"?