avatar

dcollins

Platinum Blonde

Posts: 2,365 Member Since: 27/01/2011

Lead

Feb 26 14 1:54 PM

Tags : :

In my continuing work avoidance efforts (Facebook) I’ve found a number of people, some very experienced in technical areas, that don’t really understand digital sampling.  
“It’s all stair steps!” “No information can exist between the samples” “LP’s have infinite resolution!” “Shannon and Nyqvist are why CD’s sound bad"

Et cetera.

Now comes a pair of videos so well-presented, so compelling, they are sure to not change the minds of any True Believers.

But for the rest of us:

Video Episode 1: A Digital Media Primer for Geeks

Video Episode 2: Digital Show & Tell


DC
 
Quote    Reply   
avatar

lowland

Gold Finger

Posts: 360 Member Since:21/02/2011

#1 [url]

Feb 26 14 2:14 PM

Thanks Dave - I've watched the first few minutes of video #1 and have already learned that digital is older than analogue, a surprise. I'll take in the whole show tomorrow.

Quote    Reply   
avatar

podgorny

Platinum Blonde

Posts: 2,339 Member Since:27/01/2011

#3 [url]

Feb 26 14 3:17 PM

compasspnt wrote:

I know for certain that digital only has 127 steps, whereas analogue has billions, actually infinity ones.


I'm pretty sure it's only 24 steps. 16 steps on CD.

Quote    Reply   
avatar

jaykadis

Platinum Blonde

Posts: 2,430 Member Since:24/01/2011

#5 [url]

Feb 26 14 3:35 PM

podgorny wrote:

compasspnt wrote:

I know for certain that digital only has 127 steps, whereas analogue has billions, actually infinity ones.


I'm pretty sure it's only 24 steps. 16 steps on CD.

But music only has 12 steps, so it's all good.
 

Quote    Reply   
avatar

dcollins

Platinum Blonde

Posts: 2,365 Member Since:27/01/2011

#6 [url]

Feb 26 14 9:20 PM

jaykadis wrote:

podgorny wrote:

compasspnt wrote:

I know for certain that digital only has 127 steps, whereas analogue has billions, actually infinity ones.


I'm pretty sure it's only 24 steps. 16 steps on CD.

But music only has 12 steps, so it's all good.
 

We need a 12 step program for those who think it’s stair-steps.

 

Quote    Reply   
avatar

lowland

Gold Finger

Posts: 360 Member Since:21/02/2011

#7 [url]

Feb 27 14 5:46 AM

Now watched - very good, I've posted links on Facebook. Confirmed some things I already knew and informed me in a clear way about a few things I didn't.

Quote    Reply   
avatar

cgc

Gold Finger

Posts: 944 Member Since:06/02/2011

#8 [url]

Feb 27 14 4:10 PM

I wish he had made it very clear that the mathematical equation for the analog LCR filter and the digital version are identical and only the precision of the solution differs in the two methods of implementation. The biggest misconception about analog is that it somehow is not math based.

Quote    Reply   
avatar

jerry tubb

Gold Finger

Posts: 655 Member Since:06/02/2011

#11 [url]

Mar 2 14 7:22 PM

Obviously a very bright loquacious young tech engineer, but when Monty smugly compared cassette tape to 6-bit audio, and the best analog tape to 13-bit audio I rolled my eyes. Applicable only in terms of noise floor I suppose, but not in resolution. The other thing I thought about, is that he only used tones for the testing, sine & square. Additional listening with complex music would be desirable in making judgements, but the. It can get pretty subjective. The stair step vs lollipop certainly was educational, a better perspective. Good stuff. I'd be interested in his paper on why 24-bit/192kHz isn't needed. I'm a fan of 24/96, but 192 certainly may be overkill, unless you're Neil Young };~)> JT

Quote    Reply   
avatar

lowland

Gold Finger

Posts: 360 Member Since:21/02/2011

#13 [url]

Mar 3 14 5:48 AM

Mike Kemp (designer for SADiE, Sintefex, Focusrite and JoeCo among others) responded to my Facebook posting of these videos. He said (with permission):

"Pretty good basic stuff, nice to see it demonstrated so well. I didn't think I'd watch it all, but it was interesting, and, let's face it, audio has always floated my boat, so to speak. 

A couple of points I'd mention. 

Of interest to mastering engineers is the "Gibbs effect" overshoot on "square" edges. Not may people realise that an analogue channel designed to cope with peak digital sine waves will see these larger signals on sharp edges, especially common in computer generated or edited signals. From memory I seem to recall that these peaks are up to almost 20% above the sine wave peak. Many analogue channels can distort on these before they get to analogue volume controls if programme is mastered close to the 0dBFS mark. I don't suppose that'll stop people wanting the loudest digital file possible though! 

Another point is regarding dither: It's probably true that with 16-bit audio the dither effect is inaudible on programme recorded near the peak level, as stated by the presenter. But that doesn't stop people turning up the volume on the fade-out or quiet passages and hearing the horrible distortion of undithered audio. I recall this used to show up particularly badly on sustained piano notes left to fade naturally. And it's probably one of the things that gave the first decade or so of CDs such a bad name. At that time there was a lot of badly designed gear and inexpert operators."

Plus:

"I too query the characterisation of "the best professional analogue machine with noise reduction" to "13-bits of resolution". IIRC the noise floor on our Studer 24-track was something like 70dB below operating level, and you could push to +10 over operating level with limited (but quite nice) distortion, and another 6dB or so if that was your whim (or you were stuck with VU meters!) Add 15dB of Dolby A and the s/n ratio was getting on for 100dB, which could be compared with 16-bit. And 24-track with its 1mm track width was noticeable worse than 16-track with at least twice the track width, giving another 3- 4dB of s/n. Half inch mastering gives you another few dB, enough to turn off the Dolby."

And:

"BTW, I said "At that time there was a lot of badly designed gear and inexpert operators.". Actually, I bet there is even more badly designed gear (and algorithms) about now, and even more inexpert operators. But at least the top end is now very good, and many of the best people have learnt from the mistakes of early digital recording."
 

Quote    Reply   
avatar

jerry tubb

Gold Finger

Posts: 655 Member Since:06/02/2011

#14 [url]

Mar 3 14 9:13 AM

podgorny wrote:
Jerry, I think maybe you need to watch it again. :)

Why? Please elaborate!

I may be one of those guys that Dave mentioned in the OP, or one foot in each camp

yeah, I know... }:~)>

JT

Last Edited By: jerry tubb Mar 3 14 9:32 AM. Edited 2 times.

Quote    Reply   
avatar

dcollins

Platinum Blonde

Posts: 2,365 Member Since:27/01/2011

#15 [url]

Mar 3 14 1:52 PM

As far as early digital and distorted fade-outs are concerned, it had more to do with the terrible low-level linearity performance of the early A/D’s than the lack of d*ther. Below -40 it was *showtime* for most systems.


DC

Quote    Reply   
avatar

dcollins

Platinum Blonde

Posts: 2,365 Member Since:27/01/2011

#17 [url]

Mar 3 14 5:22 PM

bob olhsson wrote:
The low level linearity may be why designer d*ther seemed useful back then.

Actually you have to have linearity _below_ the target wordlength, so until the widespread use of sigma/delta A/D’s you were never getting the full effects of d*ther at 16 bits anyway.  Although I was recently listening to some old CD’s made on 1610’s (measured performance specs rated NC-17) and it actually sounded pretty decent.  There’s a bit of crash/brashy cymbals going on, but seems better than I remember it. 

DC

Quote    Reply   
avatar

fenris

Platinum Blonde

Posts: 1,450 Member Since:26/01/2011

#18 [url]

Mar 3 14 7:27 PM

More howlers from Beardy Digital Man. (points at analog circuit board) "Analog componentry just doesn't have the flexibility of a general purpose computer. Adding a new feature to THIS beast ... yeah, it's probably not going to happen. On a digital processor, though, just write a new program." That's funny, I added a LOT of new features to my analog console, it was a lot easier to learn the workings of state variable filters etc. than it would be to teach myself programming, and the results were more musical-sounding than the latest modeling software designed by teams of quantum physicists.

Quote    Reply   
avatar

dcollins

Platinum Blonde

Posts: 2,365 Member Since:27/01/2011

#19 [url]

Mar 3 14 7:58 PM

fenris wrote:
More howlers from Beardy Digital Man. (points at analog circuit board) "Analog componentry just doesn't have the flexibility of a general purpose computer. Adding a new feature to THIS beast ... yeah, it's probably not going to happen. On a digital processor, though, just write a new program." That's funny, I added a LOT of new features to my analog console, it was a lot easier to learn the workings of state variable filters etc. than it would be to teach myself programming, and the results were more musical-sounding than the latest modeling software designed by teams of quantum physicists.

Wait, what?  Howlers?  More howlers?

Adding features to hardware can be done.  Of course. Depending on the complexity, and analog consoles are not really very complex, it can certainly be done.  But software is by its very nature changeable.  That's what it is.  

Most hardware designers are glad if there are no revisions.  I know I am.  While software is a pretty fluid thing.

Fwiw:

http://ewdijkstra.net


DC

 

Quote    Reply   
avatar

bob olhsson

Aqua Marine

Posts: 3,343 Member Since:25/01/2011

#20 [url]

Mar 3 14 9:37 PM

My experience with the old stuff is that it often sounds really good until you apply any kind of digital signal processing. I think fragile is probably the best term for it.

Quote    Reply   
Add Reply

Quick Reply

bbcode help