Gibbs Phenomenon [Archive] - Audio & Video Forums

PDA

View Full Version : Gibbs Phenomenon



manlystanley
12-05-2009, 04:24 AM
My new turntable came last night and after I hooked it up, one my sons was free, so I took him down stairs to my listening room and showed him the huge sound quality difference between CD and Vinyl. I was surprised, because he was NOT surprised.

This son is a senior in college with a double major in Physics and Math (With a full scholarship). He mentioned that one of the classes he took on Fourier Transforms had a big section on "Gibbs Phenomenon", which describes one of the weakness in modern day signal processing algorithms.

He said that while higher sampling rate can minimize this effect, that changes in dynamics are almost impossible for CD/SACD systems to be made to sound as good as Vinyl. Further higher frequencies are more problematic for CD/SACD systems.

This then explains why CD's sound 'harsh' to me AND music with more percussion sounds distorted. It's because there missing many of the harmonics.

Not sure if this is new to you, but it was me. I've attached a couple of links for your reading pleasure:

http://en.wikipedia.org/wiki/Gibbs_phenomenon

http://www.stereophile.com/features/424/index8.html

02audionoob
12-05-2009, 11:54 AM
Seems like explaining good sound quality with math is like choosing Miss Universe with it. Someone still might like Miss Ukraine better. :D

http://farm3.static.flickr.com/2778/4160353965_916a10833a_m.jpg

poppachubby
12-05-2009, 12:06 PM
You're so right.

http://www.signaltribunenewspaper.com/wp-content/uploads/2009/02/bulldog-beauty-contest-3.jpg

02audionoob
12-05-2009, 12:09 PM
Chubbs...that one made my day. I'd give you a greenie for that if I could.

manlystanley
12-05-2009, 04:54 PM
Funny. But, believe it or not, the theory behind it is actually important.

Best Regards,
Stan

02audionoob
12-05-2009, 05:42 PM
The article by Stereophile is 14 years old and makes reference to even older information...a huge length of time in digital technology. Do you have anything more current? My comment might have been said with a bit of humor, it was meant at least somewhat seriously.

atomicAdam
12-05-2009, 05:51 PM
I have to say I've heard a couple demos of CD vs Vinyl vs DVR 24bit wave files and by far - on a great system - CD sound the worst.

I can't wait till everything is made in 24bit or better. Vinyl is still king in my book, but dang if hidef audio isn't damn good as well.

I was here today with a bunch of other 'audiophiles' at JVS house with the SG2 turntable and yadda yadda - like you know $26k turntable, and a bunch of other ridiculously price gear and the PSAudio Transport and DAC I talked about in a previous thread.

The Vinyl was fantastic - HDAudio was fantastic - the CDs were irritating. Same songs - different formats. Amazing!

Death to 16bit! :cornut: :cornut:

http://gallery.audioreview.com/data/audio//500/medium/hi-fi-today.jpg

02audionoob
12-05-2009, 06:53 PM
In my experiences with quality gear, the CD format can sound terrific. My own system doesn't sound harsh with CDs...at least not to me. I think the difference is often overstated. There are certainly recordings on CD that don't sound good, but I really question the concept that digital audio is inherently flawed due to tests. I have some bad-sounding CDs, but I don't blame that on the format. I blame it on those who made the recording.

As much as I love vinyl, there are some CDs in my collection that my analog setup can't match. Maybe I could get there if I spent $10k on a turntable and cartridge...maybe not...but I've got about $1k in my CD player and DAC and I'm very pleased with it. And I'm not even a big digital proponent.

manlystanley
12-05-2009, 07:05 PM
The article by Stereophile is 14 years old and makes reference to even older information...a huge length of time in digital technology. Do you have anything more current? My comment might have been said with a bit of humor, it was meant at least somewhat seriously.

We'll the Wikipedia article is current, (but it has references to sources written in 1899), does that make it suspect?? :-)

I was just trying to return a little bit of the kindness that has been shown to me by people like yourself, chubbs, and others. I've learned a huge amount from you guys (and hope to learn much more)!

In the past when new people join this forum and can't understand why anyone would listen to old fashioned vinyl (e.g aren't CD's better?) the explanations given didn't really resonate with me. But, knowing that modern day signal processing algorithms have real problems when there are jumps in the sample signal strength which lead to "large oscillations near the jump"--really helps me understand why vinyl is better sounding.

Also, my one son was telling me there are many other reasons why vinyl sounds better, but, it was late so we went to bed.

Best Regards,
Stan

02audionoob
12-05-2009, 07:14 PM
The physics phenomenon is independent of time, but what I was getting at is the technolgy of digital audio advances continually and might not necessarily be hobbled by issues that were problematic in 1995. My Pioneer CD player from 1986 cost more than my current Music Hall if you adjust for inflation. The Music Hall absolutely destroys it. Things change.

As much as I love vinyl, I would never say across the board one format sounds better. My Diana Krall Live in Paris CD and Shelby Lynne Just A Little Lovin just might sound better in my system than any vinyl record I have. Of course, there are the limitations of my Music Hall MMF-5, my Goldring Eroica cartridge and my records.

atomicAdam
12-05-2009, 07:22 PM
The physics phenomenon is independent of time, but what I was getting at is the technolgy of digital audio advances continually and might not necessarily be hobbled by issues that were problematic in 1995. .

yeah - go hear a 24bit HDAudio - it is pretty dang awesome

manlystanley
12-06-2009, 04:58 PM
[QUOTE=02audionoob]The physics phenomenon is independent of time, but what I was getting at is the technolgy of digital audio advances continually and might not necessarily be hobbled by issues that were problematic in 1995. My Pioneer CD player from 1986 cost more than my current Music Hall if you adjust for inflation. The Music Hall absolutely destroys it. Things change.

As much as I love vinyl, I would never say across the board one format sounds better. My Diana Krall Live in Paris CD and Shelby Lynne Just A Little Lovin just might sound better in my system than any vinyl record I have. Of course, there are the limitations of my Music Hall MMF-5, my Goldring Eroica cartridge and my records.[/QUOTE

That's a good point. As technology improves and mitigation strategies are developed these challenges are slowly overcome.

Reticuli
12-07-2009, 02:53 AM
Humans cannot hear beyond 18-20bit of dynamics due to ear/brain compressing higher dynamics at extreme volumes, and even under the best & loudest circumstances can't hear over about 20-25khz. And usually you can't get both at the same time!

Most of Red Book's problem is the 44.1khz part. If CD was 16/48, few even audiophiles would be complaining. With good dacs and source material, only the smallest percentage of golden ears can hear a difference between 18 or 20bit/48khz and 24/192. If CD was 18 or 20bit and 48khz, it would have lived up to the "perfect sound forever" marketing.

The idea that the higher res formats actually have more audible detail is really nonsense. All that extra bandwidth is hogged up by ultrasonics or more accurate rendering of the noise floor. But what it does is move the unavoidable fakery band higher.

The garbage band isn't even the issue. DACs have to fudge above a certain band even below that sampling frequency's Nyquest half. To get it accurate would take infinite processing or infinite time. It comes down to the way the high frequencies are interpolated, rather than precisely determined, in the final stage of the dac instead.

CD has its most artificial "guessing" in the upper vocal band still. SACD & 48 khz pushes it up a little higher...just enough to get it out of the range the human ear is most sensitive to. 96khz further still, and 192khz into the higher treble but still in the audible band.

SACD also has a dither-like quality that makes it more "analog-like" in that your ear has difficulty latching onto a predictable pattern of digitalness, but it also adds more noise-like low-level distortion. You get better percussives, smoothness, and that lack of digital signiture, but may not have the low-level detail of even 16/48...at least that's what the math supposedly suggests. So pluses and minuses.

Also, a great many SACDs are badly mastered (Dark Side of the Moon!) or taken from PCM (what's the point?), anyway, not to mention you cannot produce or remix in DSD without converting to a multi-bit scheme along the way. All those expensive DSD consoles do it. So it's a good archival format for Columbia/Sony and for direct tape transfers & live recordings, but you wouldn't record a rockband in DSD.

As to whether what you don't like about CD is from the Gibbs phenomenon, I doubt it. That applies to very extreme harmonics only. It will occur on ANY converter at any bit depth or resolution. So even if your player is a month old and cost a fortune, it'd be incapable of making a saw or square wave perfect. Hell, a lot of very expensive AD/DA converters will mangle test waveforms much worse than that wiki shows, like unrecognizable, but sound very good with music.

You use those types of test signals on analog gear, not digital.