Memresistor & Audio

"David Grant" <[email protected]> wrote in message
news:[email protected]
> HP announced that it's beginning to manufacture the
> memresistor in 2009. Any speculation that this device
> will have a practical application in analog audio
> storage? I haven't read any specs on dynamic range, but
> assuming it beats CD quality, might we have an analog
> storage device that's as good as digital but without
> quantization error?

It appears that this device remembers the direction of current flow. Since
current can only flow in 2 directions, it would appear to be only capable of
binary (digital) memory.

>
> > HP announced that it's beginning to manufacture the
> > memresistor in 2009. Any speculation that this device
> > will have a practical application in analog audio
> > storage? I haven't read any specs on dynamic range, but
> > assuming it beats CD quality, might we have an analog
> > storage device that's as good as digital but without
> > quantization error?
>

when proper dithering applied to a digital system, quantization error
is removed, it becomes the same as analog with a noise floor, but with
a hard clipping upper limit..


Mark

"Mark" <[email protected]> wrote in message news:d41bbe42-2f0f-4c03-99eb-e9d00fd56670@79g2000hsk.googlegroups.com...

> > > HP announced that it's beginning to manufacture the
> > > memresistor in 2009. Any speculation that this device
> > > will have a practical application in analog audio
> > > storage? I haven't read any specs on dynamic range, but
> > > assuming it beats CD quality, might we have an analog
> > > storage device that's as good as digital but without
> > > quantization error?


> when proper dithering applied to a digital system,

"Proper" dithering ??

Where does "dither" enter the picture when recording or playing back ?!?

> quantization error is removed,

I'm not sure I follow you here....

> it becomes the same as analog with a noise floor,

Oh ?!?!? Where did the noise floor come from? The "proper" Dither ??

> but with a hard clipping upper limit..

Hmmm...... analogue clips quite nicely if you make it do so.


DM

On Tue, 15 Jul 2008 03:46:03 GMT, "David Morgan \(MAMS\)"
<findme@m-a-m-s.comC/Odm> wrote:

> Where did the noise floor come from? The "proper" Dither ?

Yes! Mark's line of thought was popularized here by
Bob Cain, much missed, who said that an ideal (well,
you gotta start somewhere) conversion from analog-to-
digital-to-analog (*) was exactly the same thing as
perfect, but with a small noise added.

The small noise is the dither, added to randomize
the errors in approximating. Another way of saying
it is that dither spreads the errors over a bigger
canvas, until they're just random noise.

For my money, "dither" is the coolest idea of the
last century, but if I were educated, I might feel
otherwise. People probably wrote some books and stuff...
whatever.


Much thanks, as always.
(* Oh yeah, this also presumes that bandwidth limiting
happens in the right places).

Chris Hornbeck

"Chris Hornbeck" <[email protected]> wrote in message

> For my money, "dither" is the coolest idea of the
> last century, but if I were educated, I might feel
> otherwise. People probably wrote some books and stuff...
> whatever.


Personally, I see no need what-so-ever for dither with the exception
of bit-depth conversion (and perhaps fades using poor quality software).

DM

David Morgan (MAMS) wrote:
> "Chris Hornbeck" <[email protected]> wrote in message
>
>> For my money, "dither" is the coolest idea of the
>> last century, but if I were educated, I might feel
>> otherwise. People probably wrote some books and stuff...
>> whatever.
>
>
> Personally, I see no need what-so-ever for dither with the exception
> of bit-depth conversion (and perhaps fades using poor quality software).
>
Absolutely vital. Fortunately 99.99% of the time it gets done for you by
ambient noise so you don't need to worry about it.

d

David Morgan \(MAMS\) <findme@m-a-m-s.comC/Odm> wrote:
>"Chris Hornbeck" <[email protected]> wrote in message
>
>> For my money, "dither" is the coolest idea of the
>> last century, but if I were educated, I might feel
>> otherwise. People probably wrote some books and stuff...
>> whatever.
>
>Personally, I see no need what-so-ever for dither with the exception
>of bit-depth conversion (and perhaps fades using poor quality software).

But we are ALWAYS doing bit depth conversion all the time.

In fact... that is what a fade is! It's shifting bits to the side. And
when you fade, if you have good quality software, dither is going on.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

On Thu, 17 Jul 2008 15:02:18 GMT, "David Morgan \(MAMS\)"
<findme@m-a-m-s.comC/Odm> wrote:

>Personally, I see no need what-so-ever for dither with the exception
>of bit-depth conversion (and perhaps fades using poor quality software).

I'm sure you're right; I have no reason to doubt ya!
But it's just such a cool idea: we can't have
perfection without a small penalty, and paid in a
foreign currency. Distortion is muddled with noise;
sampled time is muddled with averaged time.

Had to look this up to get it just right:
"History is facts which become lies in the end; legends
are lies which become history in the end."
-Jean Cocteau, as quoted in The Observer (22 September 1957)


Much thanks, as always,
Chris Hornbeck

On Tue, 15 Jul 2008 04:13:54 GMT, Chris Hornbeck
<[email protected]> wrote:

>On Tue, 15 Jul 2008 03:46:03 GMT, "David Morgan \(MAMS\)"
><findme@m-a-m-s.comC/Odm> wrote:
>
>> Where did the noise floor come from? The "proper" Dither ?
>
>Yes! Mark's line of thought was popularized here by
>Bob Cain, much missed, who said that an ideal (well,
>you gotta start somewhere) conversion from analog-to-
>digital-to-analog (*) was exactly the same thing as
>perfect, but with a small noise added.
>
>The small noise is the dither, added to randomize
>the errors in approximating. Another way of saying
>it is that dither spreads the errors over a bigger
>canvas, until they're just random noise.
>
>For my money, "dither" is the coolest idea of the
>last century, but if I were educated, I might feel
>otherwise. People probably wrote some books and stuff...
>whatever.

I always found this article intersting and informative (and check
out the dates of the references, the idea was around long before this
app note):
http://www.national.com/an/AN/AN-804.pdf

And that doesn't even mention "noise shaping," which (for CD/44.1k
sample rates or higher) puts most of the noise into the top octave or
two of the hearing range where the ear is much less sensitive to
low-level noise, making the overall noise sound that much lower still.

As for this "memresistor," I'm not convinced that it's the "fourth
fundamental component of electronics" nor that a physical
manifestation of it is THAT revolutionary. Presuming you could put
millions or billions on a semicondutor chip and switch an analog
signal among them at the usual 44.1k or 196k sample rates, well that
might be fun, like those Winbond audio recording chips, but we already
have a/d's d/a's and big disk drives/flash memory to hold the digital
side of things, and they're all fairly mature technology.

Sure, the memresistor will have its applications, but it's no holy
grail of electronics.

Oh, and for "quantization error" the OP was asking about, there's
two kinds of quantization in digital systems: amplitude and time. The
memresistor will be like an element in the Winbond chips, or like the
capacitor in the old BBD chips - you still have the time-wise
quantization.

And considering the consumer market has basically gone from 16-bit
PCM (CD) to that format but lossily encoded in mp3's, any "better"
quality audio device is unlikely to make it to consumer market, so it
might be marginal to make it for pro and high-end audio markets.

>
>
>Much thanks, as always.
>(* Oh yeah, this also presumes that bandwidth limiting
>happens in the right places).
>
>Chris Hornbeck