[LAD] Re: [linux-audio-dev] Direct Stream Digital / Pulse Density Modulation musing/questions

Maitland Vaughan-Turner maitlandvt at gmail.com
Tue Sep 25 03:50:02 UTC 2007

On 9/24/07, Paul Davis <paul at linuxaudiosystems.com> wrote:
> On Mon, 2007-09-24 at 13:13 -0700, Maitland Vaughan-Turner wrote:

> > oh yeah, why is that?  acoustic waves are continuous, analog
> > representations are continuous.  The more samples we can get the more
> > closely digital representation can mimic the analog which is far more
> > like the pressure waves than a series of pulses could ever be.
> there are lots of reasons why its wrong. information theory is one angle
> to take: how much information is being delivered per unit time. biology
> is another angle to take: how the human ear actually decodes acoustic
> pressure waves. non-linearities in pressure transducers (speakers etc)
> are another angle. the moment you convert an acoustic pressure wave into
> an electrical signal, its properties start to change. leaving it in
> analog form doesn't change it a lot. converting it to digital of any
> type changes the properties quite a bit, but this makes no difference if
> a symmetrical operation is possible when converting back to an analog
> electrical.
> but basically: "more pulses with less information per pulse" isn't
> equivalent to "less pulses with more information per pulse" and it
> certainly isn't equivalent to "continuously varying analog signal".

I dig what your saying, but when was the last time you listened to
just a single sample?  Granted a 24 bit sample contains a lot more
data than a 1 bit sample.  This is totally obvious.  But when you look
at a whole chunk of samples, only the first several samples of the 1
bit stream are a question mark.  After several samples it falls into
line and the amplitude can be accurately represented.

Now, I understand that 1-bit 2.8 Mhz can not achieve the dynamic range
of 24 bit samples, but it can surely represent more detailed
waveforms.  Besides, there is only one variable to maximize (instead
of two).  What happens when we turn that mega into a giga?  (I know, I
know, an even bigger processing nightmare...  hahaha)

Oh, and btw, just because I like DSD doesn't mean I don't like PCM!  I
totally dig your work, and I use Ardour all the time.  (well, ok, it's
broken on my box right now, but when it's working I use it all the
time! :)  I mean, c'mon dude, you're like a celebrity!  Thanks for
even talking to me =)


More information about the Linux-audio-dev mailing list