Digital Audio Explained (from an engineer's standpoint)

[E-Stat]

Nice work, E-Stat. You combined an ad-hominem attack, which is prohibited by AK's TOU, with a factual error, and all in one short sentence.

Let's review your comments to which I referred:

"If the difference in sound quality between hi-res digital and CD is as significant as you and other hi-res proponents claim, that $300 player should have been more than good enough to make the point."

More than good enough? Did I misstate your point in any way? And please quote what I've actually said about how "significant" the differences are.

The 8x-oversampling standalone DAC in my main system retailed for $1200 when it was new, and the CDP that I'm using as a transport carried a $350 retail price. That adds up to more than $1500, not including the jitter reducer or the two coaxial cables that are also part of my CD front end.

It would appear then that more is required to realize the benefits of digital music than a $300 player.

Did I misstate your point in any way?

 

[E-Stat]

Have you ever looked at the square wave response of a typical phono cartridge? You'd probably find the ringing on the leading edges appalling, yet you've sung the praises of analog elsewhere in this thread.

Speculation based on different criteria so noted.

 

[JP]

I never put words in anyone's mouth. Going back and reviewing the thread, the conversation follows the inference that digital copies suffer degradation. I said bits are bits, and you called that a theory.

Bias and psychoacoustics aside, of course when you look at an entire reproduction chain there are tolerances and differences, audible or not. But that wasn't the indicated argument in your comment of bits being bits a theory, nor was it the discussion Allen had.

If Allen gave the production house a CD (probably not) or CD image to produce, the production should have been bit perfect with the source he gave them. If it wasn't, there was an error or they changed something, and if that's the case one can't say that the difference he claims to hear was caused by generational degradation of digital.

As for copies being bit perfect, aside from the simple fact that this "phenomena" doesn't exist anywhere else in the digital data world aside from audio, I've tested it. We did 10,000 generational copies of a wav file. All 10,000 copies were bit perfect, and predictably taking the first and last or random samples for ABX, no one could hear a difference.

With the proper test equipment it's very easy to check a CD for data errors (both soft and hard), to measure jitter in a bit stream, etc. While there's lots that can go wrong, it seldom does, particularly with competently engineered components.

The chances of the same reproduction chain manifesting audible errors consistently between two different but bit identical CDs that can be read without unrecoverable error in a repeatable and predictable fashion are none to none.

If I'm misunderstanding the position your advocating, please let me know.

Oh, and on the high res vs. Redbook argument, it's really simple. Take a high res file, convert it to Redbook (level matched and all that jazz), and ABX them. I've a friend that's done this with dozens of people - no one can hear a difference. Dead simple for you to try yourself.

 

[E-Stat]

I never put words in anyone's mouth.

Except of course your commentary about "bit perfect" copies. Allen didn't say that. I didn't say that. Pray tell, who exactly said that in this thread?

I said bits are bits, and you called that a theory.

I really wish folks would actually quote what I say when they respond to me. That would provide context. Let's look at what I said:

Bits are not bits and there's far more to go wrong in the real world.

If that were true, the DVD player I got free from Omaha Steaks would sound just as good as an EMM Labs or DCC player. A lot happens between the bits and the analog output.

If Allen gave the production house a CD (probably not) or CD image to produce, the production should have been bit perfect with the source he gave them. If it wasn't, there was an error or they changed something, and if that's the case one can't say that the difference he claims to hear was caused by generational degradation of digital.

Speculation so noted.

As for copies being bit perfect, aside from the simple fact that this "phenomena" doesn't exist anywhere else in the digital data world aside from audio, I've tested it. We did 10,000 generational copies of a wav file. All 10,000 copies were bit perfect, and predictably taking the first and last or random samples for ABX, no one could hear a difference.

Do tell us of your methodology that "we" did.

Oh, and on the high res vs. Redbook argument, it's really simple. Take a high res file, convert it to Redbook (level matched and all that jazz), and ABX them. I've a friend that's done this with dozens of people - no one can hear a difference. Dead simple for you to try yourself.

And engineers have compared live feeds with various resolution results and come to the exact opposite conclusion. Maybe its your ABX box!

 

[JoeESP9]

If "bits is bits" and there is a question as to the truthfulness of this it's quite easy to find out for yourself. Get two CD-RW disks. copy an original CD to one and then copy it to the second and then back to the first etc. Ten generations should be sufficient to either prove or disprove the issue. Anyone who has any questions can try this him/her self.

 

[hwirt]

In the original thread, I asked if Monty ever addressed the inevitable ringing resulting from steep anti-aliasing filters. Never got a response. Do you know?

Showing test tones on an o-scope is kinda cool, but completely misses the point of why sampling at higher rates is desirable. Which Allen did mention. No steep filters required to ensure Nyquist is satisfied to capture 20k.

Steep analogue brick wall filters have been obsolete for thirty years or more, Phillips used oversampling and much more gentle digital filters in the early 80's for redbook audio (16/44). It's no longer an issue and hasn't been for decades.

Review Monty's video once again. start watching as time interval 4:00. In this section of the video he demonstrates that the digitally processed signal output is a perfect sine wave. He increases the frequency 16, then 20 KHz. The sine wave is still perfectly sinusoidal, showing no distortion products from ringing or any other anomalies.

 

[E-Stat]

Steep analogue brick wall filters have been obsolete for thirty years or more,

Absolutely. The steep brickwall filters are now in the digital domain. The Nyquist requirement, however, persists. Please return to post # 27 where I made that observation.

Review Monty's video once again. start watching as time interval 4:00. In this section of the video he demonstrates that the digitally processed signal output is a perfect sine wave. He increases the frequency 16, then 20 KHz. The sine wave is still perfectly sinusoidal, showing no distortion products from ringing or any other anomalies.

Yeah, that's great with a test signal on the o-scope. Up until the point the necessary digital anti-aliasing (aka brickwall) filter is applied to the signal for storage and playback.

 

[hwirt]

Absolutely. The steep brickwall filters are now in the digital domain. The Nyquist requirement, however, persists. Please return to post # 27 where I made that observation.


Yeah, that's great with a test signal on the o-scope. Up until the point the necessary digital anti-aliasing (aka brickwall) filter is applied to the signal for storage and playback.

The filters are no longer brick wall due to oversampling, stop beating a dead horse. The DAC outputs a perfect sine wave at 20 KHz, watch the video.

"The term oversampling is also used to denote a process used in the reconstruction phase of digital-to-analog conversion, in which an intermediate high sampling rate is used between the digital input and the analogue output. Here, samples are interpolated in the digital domain to add additional samples in between, thereby converting the data to a higher sample rate, which is a form of upsampling. When the resulting higher-rate samples are converted to analog, a less complex/expensive analog low pass filter is required to remove the high-frequency content."

 

[dumptruck]

Absolutely. I suspect most every one. It was an expensive venture at first to introduce new recording standards that are now universally accepted.

You suspect most every one did a successful ABX test, and yet not one of them actually used this never-seen bit of data in their sales pitch? How strange.

P.S. You mentioned recording here and again later in your post. Nobody in their right mind would recommend recording straight to redbook as ideal. The minimums for recording are necessarily greater than for playback, whatever the specific format you want to argue for. I think this has been noted several times, but if we're going to keep mentioning mentioning both recording and end user playback, we have to keep saying which one we mean.

 

[E-Stat]

The filters are no longer brick wall due to oversampling, stop beating a dead horse. The DAC outputs a perfect sine wave at 20 KHz, watch the video.

They still ring in the real world at 44k even when oversampled.

Redbook filter compromises

 

[hwirt]

They still ring in the real world at 44k even when oversampled.

http://www.ayre.com/white_papers/Ayre_MP_White_Paper.pdf

Ok, so it's been corrected using an oversampling filter. Like I said.... Thank's for proving my point.

 

[E-Stat]

You suspect most every one did a successful ABX test, and yet not one of them actually used this never-seen bit of data in their sales pitch? How strange.

Ask them. Aside from Harman who does blind testing on speakers, I've never seen such used in advertising. If all amplifiers sound the same, then why don't you find companies that produce $300 amplifiers tout their success against $30,000 amplifiers? Your guess is as good as mine.

P.S. You mentioned recording here and again later in your post. Nobody in their right mind would recommend recording straight to redbook as ideal. The minimums for recording are necessarily greater than for playback, whatever the specific format you want to argue for. I think this has been noted several times, but if we're going to keep mentioning mentioning both recording and end user playback, we have to keep saying which one we mean.

You mean there are losses in digital copies? Oooooohhhhh.

 

[E-Stat]

Show me an example, not a sales brochure.

Ignore the text and view the graphs.

Tell me of a company who produces high performance players or DACs that doesn't make their own filters? PS Audio talks of their "hand crafted" filters. The Gamut CD-1 I used to own used a similar "listen" compromise filter with their player that measured a bit down at 20 khz but retained better phase response. They oversampled, too.

Sheesh! If Redbook is "perfect sound forever" then by all means enjoy it!

 

[hwirt]

Ignore the text and view the graphs.

Tell me of a company who produces high performance players or DACs that doesn't make their own filters? PS Audio talks of their "hand crafted" filters. The Gamut CD-1 I used to own used a similar "listen" compromise filter with their player that measured a bit down at 20 khz but retained better phase response. They oversampled, too.

Sheesh! If Redbook is "perfect sound forever" then by all means enjoy it!

Ayer claims to have corrected for the effects of brick wall filtering using their oversampling technique. So have many others, and have been doing it for years.

 

[dumptruck]

Ask them. Aside from Harman who does blind testing on speakers, I've never seen such used in advertising. If all amplifiers sound the same, then why don't you find companies that produce $300 amplifiers tout their success against $30,000 amplifiers? Your guess is as good as mine.

Well, I don't think all amplifiers always sound the same (easy to prove if you choose the right ones and the right speakers), but regardless it's pretty difficult to do a bulletproof ABX with amplifiers and nobody wants to get in legal trouble. However, a company doing an ABX that shows someone can tell the difference between their own new higher quality digital files when played back at full quality and when played back after downsampling, that's an easy test, and involves no other company's products. Until somebody shows me otherwise, I'll continue to assume nobody does it because nobody passes.

You mean there are losses in digital copies? Oooooohhhhh.

No, I mean there are losses (to use the term loosely) in digital manipulation and mixing, which you want to be able to do when you're producing a record.

 

[E-Stat]

Ok, so it's been corrected using an oversampling filter. Like I said.... Thank's for proving my point.

"Corrected" with a few compromises.

The FR rolls of before 20k and there is still one cycle of post ringing. Neither compromise is required with a sufficiently high sample rate. Ask Charles Hansen why he produces SACD players. Or refer to his post on another site I referenced earlier in #30 to get his opinion.

 

[hwirt]

Double post deleted.

 

[hwirt]

"Corrected" with a few compromises.

The FR rolls of before 20k and there is still one cycle of post ringing. Neither compromise is required with a sufficiently high sample rate.

Not much of a compromise don't you think. But yes higher sampling rates and bit depth is a plus, but Allan's claims of 196 KHz sounding so much better than 96 KHz is not credible. I own quite a few SACD's but not a big fan of that format, my preference would be 24/96 PCM.

 

[E-Stat]

Not much of a compromise don't you think.

But audible. Does the difference exist? Yes. Does it matter to you? Perhaps not.

But yes higher sampling rates and bit depth is a plus, but Allan's claims of 196 KHz sounding so much better than 96 KHz is not credible.

Surprisingly enough, I'm going to agree with you!

I find the most bang-for-the-buck occurs with 24/96 over 16/44. The benefits above that are definitely harder to discern. I find that 24/96 effectively eliminates the sterile sound that Redbook demonstrates and restores the ability to convey a realistic sense of "air" at the top.

 

[hwirt]

But audible. Does the difference exist? Yes. Does it matter to you? Perhaps not.


Surprisingly enough, I'm going to agree with you!

I find the most bang-for-the-buck occurs with 24/96 over 16/44. The benefits above that are definitely harder to discern. I find that 24/96 effectively eliminates the sterile sound that Redbook demonstrates and restores the ability to convey a realistic sense of "air" at the top.

I agree, 24/96 is likely the sweet spot. Very dynamic, clean and virtually no noise.