Exploring Myths in Digital Audio Part 1

[gene]

<P><SPAN lang=EN-AU style="mso-bidi-font-family: Arial"><FONT size=2><FONT face=Arial><A href="http://www.audioholics.com/techtips/specsformats/DigitalAudio.php"><IMG style="WIDTH: 125px; HEIGHT: 52px" alt=[digital] hspace=10 src="http://www.audioholics.com/news/thumbs/digital_th.gif" align=left border=0></A>Over the years myths have brewed about the limitations of digital audio and how it couldn’t possibly reproduce an analog music signal as faithfully as vintage Vinyl.<SPAN style="mso-spacerun: yes">&nbsp; </SPAN>Even experienced and well-respected audio engineers often subscribe to some of these myths. This article explores some common myths and misconceptions surrounding digital audio, as well as puts forth a more reality based perspective about this topic. </FONT></FONT></SPAN></P>
<P><SPAN lang=EN-AU style="mso-bidi-font-family: Arial"><FONT size=2><FONT face=Arial>[Read the Article]</FONT></FONT></SPAN></P>

 

[tbewick]

I thought that was quite an interesting article. My personal 'proof' that digital audio is audibly transparent was when I made a minidisc digital copy of a excellent sounding l.p. - Star Trek - The Motion Picture, by Jerry Goldsmith. This wasn't as a test for the sake of doing a test, but simply because vinyl is clumsier to play than minidisc. It sounded just as good as the original, and this was using a cheap, portable minidisc recorder/player, which would obviously be of far lower quality (ADC, jitter, etc) than studio equipment. I'd recommend anyone else to try something like this, to prove to themselves that digital audio is very good and transparent.

 

[WmAx]

<P><SPAN lang=EN-AU style="mso-bidi-font-family: Arial"><FONT size=2><FONT face=Arial><A href="http://www.audioholics.com/techtips/specsformats/DigitalAudio.php"><IMG style="WIDTH: 125px; HEIGHT: 52px" alt=[digital] hspace=10 src="http://www.audioholics.com/news/thumbs/digital_th.gif" align=left border=0></A>Over the years myths have brewed about the limitations of digital audio and how it couldn’t possibly reproduce an analog music signal as faithfully as vintage Vinyl.<SPAN style="mso-spacerun: yes">&nbsp; </SPAN>Even experienced and well-respected audio engineers often subscribe to some of these myths. This article explores some common myths and misconceptions surrounding digital audio, as well as puts forth a more reality based perspective about this topic. </FONT></FONT></SPAN></P>
<P><SPAN lang=EN-AU style="mso-bidi-font-family: Arial"><FONT size=2><FONT face=Arial>[Read the Article]</FONT></FONT></SPAN></P>

It could also be argued that the frequency components above 20kHz for sawtooth and sinewaves are not important, since we don’t hear them, so arguably what our ears are hearing are represented accurately in Figures 6-8.

One can get high frequency distortions which are normally not audible to become audible with special test signals that are in no way represenative of real music. But the distortions shown here are so subtle, and are made under such artificial conditions, that I wonder if you should have even provided the graphics. Please refer to *.

This is perhaps true, but the counter argument is that if a digital player is not playing back a 0dB 19997kHz sawtooth (or a 0dB 10kHz square wave) with even amplitude accuracy (let alone harmonic accuracy), then that represents a kind of “distortion” that is audible. By the way, analog circuits have no problems playing back sawtooth and square waves with much better amplitude and harmonic accuracy than digital.

Analog playback mediums such as vinyl have far more technical problems than a digital system. Noise, wear, high levels of harmonic distortion(in the single digits) and a very poor linearity after the complete transfer process, when comparing original master tape to final played-back record.

Many people have commented that CDs do not seem to reproduce high frequencies as well as analog sources such as vinyl and magnetic tape. Perhaps the above could at least partially explain the subjective impressions? This would also suggest one benefit of sampling at higher rates such as 96kHz or 192kHz even though our ears cannot hear past 20kHz. A higher sampling rate can help preserve amplitude and harmonic accuracy for non-sine waves at high frequencies.

The people making such comments do not establish acceptable listening test controls. The credibility of such commentary in this condition is about as believable as cable sound claims.

*If you want higher accuracy of amplitude tracking for the unusual conditions presented here(rendering a 19997 cps triangle wave with software), there is no reason to have playback equipment that has a higher sample rate. Just change the rendering process to adapt or use software with better predicitve rendering for the target sample. For Adobe Audition, render at 192kHz, apply a high resolution FFT filter(or automated filter if it is present in the software) to the signal to cut >21kHz, then downsample to 44.1kHz. See the result in figure [1]. The amplitude error as illustrated in your article will almost be completely eliminated. For actual recording, the amplitude deviations should depend on the anti-alias filter used on the ADC. I.E.; if you remove the spectra >22kHz before it is sampled, there are no components to cause such error(s), becuase you have a a signal that consists of no signals >22kHz, only the sine wave components <22kHz will be present.

-Chris

[1] http://www.linaeum.com/images/triangle_processed.gif

 

[Steve1000]

I've got my whole LP collection on minidisc. Sounds just like LPs. At first I used a Sony R50 (which eventually died of over-use) and then I used a home component CD deck. They are recorded in ATRAC 4.0 and ATRAC R -- at a bitrate of about 256 to 290 kbps. What you need to get an LP right is a lot less than that, IMHO, because the fidelity inherent in the LP medium is in fact so relatively low (as far as distortion, s/n ratio, frequency response, etc.) as compared to an essentially transparent meduim (e.g., CDs) in an absolute sense, although good LP sound is good enough to be extremely enjoyable. LPs (and minidiscs of LPs) can be more euphonic than CDs, because the sound was engineered to be very pleasant to the ear to overcome the limitations of the LP medium, IMHO. I consider the LP on minidisc experience to be more hi-fi and more enjoyable than the straight LP experience, because in making the LP recordings I had the volume turned down to get rid of any feedback, I monitored the balance where my cartridges weren't quite perfect, and I re-recorded tracks, cleaned the record, re-set the tracking force, etc. where I ran into problems on a specific record. And I get direct~random access with a remote control like a CD and titling, etc., so the end experience is really much more about the music and less about the medium. I'm left to concentrate on the music. I love hearing the sound of the original LPs as I fondly remember it... makes me feel young again.

But anyway, yeah, to hear just how transparent the digital medium can be, try making a digital recording of an LP. I first found it to be a pretty remarkable experience, hearing the digital medium reproduce the "analog" sound so effortlessly and with such transparency.

I thought that was quite an interesting article. My personal 'proof' that digital audio is audibly transparent was when I made a minidisc digital copy of a excellent sounding l.p. - Star Trek - The Motion Picture, by Jerry Goldsmith. This wasn't as a test for the sake of doing a test, but simply because vinyl is clumsier to play than minidisc. It sounded just as good as the original, and this was using a cheap, portable minidisc recorder/player, which would obviously be of far lower quality (ADC, jitter, etc) than studio equipment. I'd recommend anyone else to try something like this, to prove to themselves that digital audio is very good and transparent.

 

[mtrycrafts]

<P><SPAN lang=EN-AU style="mso-bidi-font-family: Arial"><FONT size=2><FONT face=Arial><A href="http://www.audioholics.com/techtips/specsformats/DigitalAudio.php"><IMG style="WIDTH: 125px; HEIGHT: 52px" alt=[digital] hspace=10 src="http://www.audioholics.com/news/thumbs/digital_th.gif" align=left border=0></A>Over the years myths have brewed about the limitations of digital audio and how it couldn’t possibly reproduce an analog music signal as faithfully as vintage Vinyl.<SPAN style="mso-spacerun: yes">&nbsp; </SPAN>Even experienced and well-respected audio engineers often subscribe to some of these myths. This article explores some common myths and misconceptions surrounding digital audio, as well as puts forth a more reality based perspective about this topic. </FONT></FONT></SPAN></P>
<P><SPAN lang=EN-AU style="mso-bidi-font-family: Arial"><FONT size=2><FONT face=Arial>[Read the Article]</FONT></FONT></SPAN></P>

Thanks for another interesting work.
J. Stewart of Meridian, I believe their chief designers maybe, mentions the woeful shortcoming of vinyl.
Nothing will convince the true believers.

 

[Buckeyefan 1]

Thanks for another interesting work.
J. Stewart of Meridian, I believe their chief designers maybe, mentions the woeful shortcoming of vinyl.
Nothing will convince the true believers.

You'd be amazed at the sound available from a Winamp Plug-in. Check out some of these - they bring new meaning to digital processing. WmAx may enjoy a few of them.

http://www.winamp.com/plugins/details.php?id=143170

http://www.winamp.com/plugins/details.php?id=83194

This one will permanently put your vinyl in the attic
http://www.winamp.com/plugins/details.php?id=39214

 

[Chuck V]

Hello.

I’m trying to dig in deep on DAC’s and ADC’s, and the technology behind.

The thing this article takes up is this sawtooth waves and indicates that music contains lots of stuff that (over)sampling can’t take care of so to speak. A shortcoming of the CD-format from what I can understand reading the article.

How big of an issue is this in reality and is it in the frequency range that we can hear, sub 20 kHz? The article does not quantify this in any way and I have not heard or read about this anywhere else.

Inputs appreciated.

Regards
/Chuck

 

[fmw]

I've got my whole LP collection on minidisc. Sounds just like LPs.

I hear you. This evening I decided to rip my copy of JT (James Taylor) to my computer. Since it is an SACD it isn't that easy. The computer drive can't read it. The SACD player can but it won't send the data to a digital output. So I recorded it from an analog output to an old pro audio Marantz CD recorder I have. The resulting CD sounded exactly like the original SACD (in stereo.) To make sure, I administered a blind test to my wife and she couldn't tell one from the other.

So not only is digital to analog conversion effective, it is apparently pretty simple. I took a digital original, converted it to analog, converted it back to digital and recorded it, losing nothing audible in the translation. Even simple, inexpensive converters seem to do the job with amazing accuracy.

My digital dubs of vinyl records sound just like vinyl records too.

 

[westcott]

Glad to see someone who is not afraid to tackle the political topics.

I embraced digital audio a long time ago and as an engineer, I can appreciate how good the technology really is. Some smart folks in this world who have made quality music reproduction affordable and very accurate for the masses.

If there is a difference the human ear can discern, I am not one of them. I will save my thousands of dollars and stick with digital. Thank you very much.

Here is another article concerning THD some people may be interested in.

"And on Monday, July 23, National Semiconductor announced a new family of high voltage operational-amplifier (op-amp) chips that can achieve just 0.00003% THD+ noise. That means that with a full-scale signal, the combination of noise and distortion lies an extraordinary 130dB below the signal! "

THD Wars Over?

 

[Claudio53]

I bought a SACD player two months ago (my CD player was not working anymore), not a very expensive unit, a Marantz SA15S1.
Well, even if I am 54 yers old and my hearing is not what it was when I was young, I am pretty sure that there is an audible difference between CD and SACD. Especially the dynamic range is impressive, when the music stops you can really follow a smooth reverberation decay to silence. With CDs the tail of the reverberation is jaggy (I do not know if this is the right definition, exscuse me, my english is poor).
High frequencies too are better in my opinion (obviously they depend on the quality of the recording too), as is positioning in space (maybe less phase distortion effects from the anti aliasing filters?).
I am sad that not many titles are avalaible in SACD format, and that the format may soon become obsolete.

 

[fmw]

I bought a SACD player two months ago (my CD player was not working anymore), not a very expensive unit, a Marantz SA15S1.
Well, even if I am 54 yers old and my hearing is not what it was when I was young, I am pretty sure that there is an audible difference between CD and SACD. Especially the dynamic range is impressive, when the music stops you can really follow a smooth reverberation decay to silence. With CDs the tail of the reverberation is jaggy (I do not know if this is the right definition, exscuse me, my english is poor).
High frequencies too are better in my opinion (obviously they depend on the quality of the recording too), as is positioning in space (maybe less phase distortion effects from the anti aliasing filters?).
I am sad that not many titles are avalaible in SACD format, and that the format may soon become obsolete.

No question about it. You misunderstood my post. I was comparing an original SACD played on a 2 channel system with a digital to analog to digital dub played on the same system. No difference in sound at all. I was just commenting that the conversion process is very accurate and consistent.

 

[Claudio53]

No question about it. You misunderstood my post. I was comparing an original SACD played on a 2 channel system with a digital to analog to digital dub played on the same system. No difference in sound at all. I was just commenting that the conversion process is very accurate and consistent.

Sorry for the misunderstanding. I usually listen to SACD in stereo (my player by the way decodes only stereo DSD and not multi channel DSD), so I assumed that you were referring to DSD stereo.

 

[OttoMatic]

Hi, fmw,

... a digital to analog to digital dub played on the same system. No difference in sound at all. I was just commenting that the conversion process is very accurate and consistent.

Well, I'd bet there's some degradation there. There simply has to be. You couldn't do that 100 times and maintain resolution.

 

[Joe Schmoe]

Ever since they first came out, I have felt that CDs not only sounded better than LPs, but dramatically better. Unfortunately, the potential of CD has recently been undermined by the "loudness wars".

 

[fmw]

Hi, fmw,



Well, I'd bet there's some degradation there. There simply has to be. You couldn't do that 100 times and maintain resolution.

I only did it once. There could be some degradation but it isn't audible in a blind listening test so it doesn't matter.

 

[OttoMatic]

Sure. I'm not trying to argue that vinyl is better or not. But I believe that while DSP theory, Nyquist and Fourier all come together to indicate that we can "perfectly" recreate the original analog signal, it's not always that way in practice.

DACs can sound different from product to product. Similarly, ADCs must be different from product to product. Therefore, if you go through the ADA process even once, you've changed the signal. I think it would be audible. If you did it 100x, I would bet good money that my wife could hear the difference. Yeah, 1x vs 100x is a big difference, but the point is that there is a change there. Not necessarily because the underlying signal processing theory is invalid, but because of the imperfect implementation that theory.

If you can't hear a difference, there may indeed be no difference. On the other hand, perhaps the source material wasn't high enough resolution to begin with, or perhaps the overall system isn't.

Again, I'm not saying that analog is better. I'm just saying that theory vs. reality is frequently different.

 

[Starmax]

Claudio, you said your English was poor? Could have fooled me!

 

[fmw]

Sure. I'm not trying to argue that vinyl is better or not. But I believe that while DSP theory, Nyquist and Fourier all come together to indicate that we can "perfectly" recreate the original analog signal, it's not always that way in practice.

DACs can sound different from product to product. Similarly, ADCs must be different from product to product. Therefore, if you go through the ADA process even once, you've changed the signal. I think it would be audible. If you did it 100x, I would bet good money that my wife could hear the difference. Yeah, 1x vs 100x is a big difference, but the point is that there is a change there. Not necessarily because the underlying signal processing theory is invalid, but because of the imperfect implementation that theory.

If you can't hear a difference, there may indeed be no difference. On the other hand, perhaps the source material wasn't high enough resolution to begin with, or perhaps the overall system isn't.

Again, I'm not saying that analog is better. I'm just saying that theory vs. reality is frequently different.

Again, I don't doubt that there is a difference - just not an audible one. The equipment isn't too bad. I dubbed from a Pioneer Elite player through my Mackie digital mixer to a 16 track Fostex hard disk recorder. It has really good ADC's. I usually master to a Panasonic 3800 DAT but this time I simply burned it to a Marantz CD recorder - digital to digital. No EQ or compression or expansion or effects. Just a straight-through dub. The result was just dandy.

 

[OttoMatic]

Gotcha, fmw! And no problem. I do agree that the differences could be minute. Sounds like were using good equipment, and I'm sure that helped. I do admit that I've never had cause to do that type of thing myself, so I have no first hand experience.

Have a good one.

 

[tbewick]

I’m trying to dig in deep on DAC’s and ADC’s, and the technology behind.

Malcolm Hawksford's publications list at the University of Essex http://www.essex.ac.uk/ESE/research/audio_lab/malcolms_publications.html

http://citeseer.ist.psu.edu/118119.html
"Quantization Noise in Delta-Sigma A/D Converters", (1995), Robert Gray.

www.eurasip.org/content/Eusipco/1996/paper/fi_4.pdf
"Nonlinear dynamics of bandpass sigma-delta modulation". Orla Feely and David Fitzgerald.

Martin Snelgrove's web site:
http://www.dissonance.com/

http://www.elec.qmul.ac.uk/people/josh/publications.htm
Number of papers on sigma-delta modulators. Josh Reiss, Queen Mary, University of London.

http://www.eecs.harvard.edu/vlsi/techreports/tech.html
Harvard University electrical engineering website

Scala Technology Ltd:
www.scalatech.co.uk

www.grimmaudio.com/whitepapers/discrete ad converter.pdf
Bruno Putzeys (2003). 'Design techniques for high-performance discrete a/d converters'.

http://sscs.org/jssc/00bp.pdf
Ichiro Fujimori, Akihiko Nogi, and Tetsuro Sugimoto (2000). 'A Multibit Delta-Sigma Audio DAC with 120-dB Dynamic Range'.

www.utdallas.edu/~hellums/docs/EE7326/Brandt1.pdf
"Second-order sigma-delta modulation for digital-audio signal acquisition", (1991), Brian Brandt, Drew Wingard, and Bruce Wooley.

I found Professor Hawksford's 'Introduction to Digital Audio' paper quite good. I haven't read most of the other papers.

The thing this article takes up is this sawtooth waves and indicates that music contains lots of stuff that (over)sampling can’t take care of so to speak. A shortcoming of the CD-format from what I can understand reading the article.

How big of an issue is this in reality and is it in the frequency range that we can hear, sub 20 kHz? The article does not quantify this in any way and I have not heard or read about this anywhere else.

A paper by Bernd Theiss and Malcolm Hawksford, 'Phantom Source Perception in 24 bit at 96 kHz Digital Audio', included a listening test which suggested that 96 kHz, 24 bit audio might be audibly superior than 16 bit, 48 kHz audio. A paper by Julian Dunn looked at filters used in digital converters, concluding that 'it may be difficult to distinguish any beneficial effects of an increase in sampling frequency from the different filter behaviour. This should be considered when making comparisons between different rates.'

Julian Dunn - `The benefits of 96 kHz sampling rate formats for those who cannot hear above 20 kHz' Preprint 4734, presented at the 104th AES Convention, Amsterdam, May 1998, available at:
http://www.nanophon.com/