So why is decimation used at all if sample rate conversion (SRC) with
integer multiples is always possible using two steps?

For example, one could take a PCM 96000 sample rate recording, and
multiply by 147 and then divide by 320 to get to a 44100 SR.  Or
take a 192000 SR recording and multiply by 147 and then divide by 640.

I don't see how DSD provides any advantages over PCM for SRC if you
make the conversion in two steps, since the exact same SRC can also
be accomplished using PCM.

Naturally, a two-step SRC without decimation will require more time
and memory to compute - perhaps making it less suitable for
cost-effective real-time hardware implementation.  But hardware is
getting faster and more inexpensive, so I'm not even sure that is a

Is it the case that every non-real-time SRC (ie. DAW) is done in
two steps without decimation?

Again, I still don't see the advantage of DSD over PCM for SRC (I'm
not talking about word-length, only SRC).

Eric Jacobs
The Audio Archive

-----Original Message-----
From: Association for Recorded Sound Discussion List
[mailto:[log in to unmask]]On Behalf Of Konrad Strauss
Sent: Tuesday, February 22, 2005 3:35 PM
To: [log in to unmask]
Subject: Re: [ARSCLIST] software for .wav files

on 2/22/05 12:41 PM, Eric Jacobs at [log in to unmask] wrote:

> The DSD sample rate is 2.8224 MHz.  This is 64 x 44100.  However, 48000
> not divide into the DSD sample rate at an integer multiple.  Maybe I'm
> something?

It's a 2 step process. Multiply by 5 and divide by 294.

Konrad Strauss
Director of Recording Arts
Associate Professor of Music
Indiana University School of Music