naturerecordists
[Top] [All Lists]

Re: Sample Rate Conversion

Subject: Re: Sample Rate Conversion
From: "Rich Peet" <>
Date: Mon, 10 May 2004 19:07:20 -0000
Audition converts files that are opened to 32bit floating for all
edits unless you change the default to defeat it.  There is no change
made on sample rate.

I guess I don't understand the benefit of conversion to 192khz
inorder to make a "big" jump.  I thought math was math to a
computer.  I guess I will catch up on that as I record more and find
myself converting more.=20

Then there is the million dither choices to make on down sampling
which is a separate nightmare of a choice.

Rich

--- In  Walter Knapp <>
wrote:
> From: Lang Elliott <>
> >
> > Walt:
> >
> > Yes, you found a way to make it work. Both the Peak and Soundhack
> > conversions now sound virtually identical to the original file,
at least to
> > my ear.
> >
> > The magic incantation works.
> >
> > Lang
>
> Since it's your original, I'll take you as the authority on how
well it
> works. To my ear they seem to fix the problem.
>
> I'd not expect it to work as well as something programmed properly,
it
> was surprising just how well it did. It's only crudely doing
something
> that proper programming would do.
>
> Since no one has come up with the magic incantation, here it is:
>
> In Peak (or Soundhack)
>
> 1. Open the original 48khz 16bit file and save it as a 48khz 32bit
file.
>
> 2. Open that file, resample it to 192khz, then resample again, to
44.1khz.
>
> 3. Save the file as 44.1khz 16bit.
>
> This all improves the situation because it appears the two programs
do
> their math on the original samples, thus introducing some error. By
> contrast Spark XL has a settable scratch file it works on and I
have it
> at 32bit floating point. Peak and Soundhack will only do 32bit
linear max.
>
> The resampling to 192khz is that it's a lot harder to accurately
> resample to a close frequency than a greatly different one. The big
jump
> to 192khz and then the big jump to 44.1khz can be done more
accurately
> than the single tiny jump between 48khz and 44khz by these
programs. I
> don't know for sure how Spark XL (or Audition) is handling this
problem.
> Obviously they are doing it a different way.
>
> For Lang, this set of steps does not look possible in a single pass
in
> Peak's Batch processor unless there is a way to do the bit depth
without
> saving to a new file. It could probably be set up using applescript
or
> Apple Events in Filemaker. Not something I know well enough. If the
> double resampling sounds good enough that part can be set up in the
> batch process.
>
> Walt
> 



________________________________________________________________________
________________________________________________________________________

<Prev in Thread] Current Thread [Next in Thread>
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU