The Four Truths of Digital
JPH-22 -- Wed, 10/29/2008 - 15:22
The four truths of digital:
Truth One: DSD is really a 4-8 bit recording system, not 1. This means it gets decimated and oversampled - two things the scheme was trying to avoid in the first place.
Truth Two: Most 16-bit "audiophile" CDs contain 20-bit level resolution and have since 1993.
Truth Three: The limit of human hearing is 20-bit level resolution.
Truth Four: Digital is hundreds of times lower in distortion vs. analog. But its distortions are spread-out spectrally and are harder to pin down. They are removing these, at last, in the latest playback gear..........
JPH
jph-22 wrote:
Truth Two: Most 16-bit "audiophile" CDs contain *20-bit* resolution and have since 1993.
What is it with you and this falacy? It doesn't matter how often the record companies marketing department (or you) repeats it, Red Book CD is *not* a 20-bit resolution format. The data is stored--and retrieved from--the discs in 16-bit words. It doesn't matter how many bits are used in the master, the data pulled from the disc has only 16 bits.
What you're claiming is akin to claiming a black and white TV program is actually in colour because it was recently re-recorded onto a medium that supports colour.
Please desist from posting this misinformation. (In fact, given that you've been called out on this at least once in the past, I'm *this* close to calling you a troll...)
You still don't know.....
Like I said before, look at the db specs of your (audiophile) recordings or ask the recording engineer himself what they are. Or read the editor of Stereophile, John Atkinson, in his "As We See it" column from July 1994 titled "Squeezing the Music". Here, Atkinson reveals the coming world of 20-bit CD sound thanks to "re-dithering" units from Meridian and Apogee.
*Until you do these things, you'll continue to be wrong on this issue.
This extended resolution is one reason why CD rivals "high resolution" systems like SACD. But ocassionally, a reviewer will prefer CD !! Like the latest AR CD player review by Jimmy Hughes in HI-Fi News (U.K.)......
JPH
I have to weigh-in on the side of RichTeer on this one. The CD can store 16-bit data. Although re-dithering systems are an improvement over creating 16-bit words from 16-bit masters, or from truncating 20-bit masters to 16 bits, the fact remains that the LSB in a 16-bit word cannot contain five bits of information.
I'm referring to the resolution/noise floor - not the actual bits. Just ask Bob Katz or Bob Stuart...or any audiophile recording engineer (from Telarc, Chesky, etc.).
With re-dithering, there is a drop of at least 20 db - on CD. It's right in the specs.
Why would the editor of Stereophile report this if it weren't true ?
JPH
Perhaps I'm misinformed, but my understanding is that a 16-bit word length can encode 65,536 discrete quantization steps. A 20-bit word length can encode 1,048,576 discrete quantization steps. I don't understand how a 16-bit word can encode more than 65,536 quantization steps.
Here's a thought experiment. Toggle the LSB of a 20-bit word on and off and then re-dither that signal to 16 bits. Would the information in the 20-bit word's LSB appear in the 16-bit words?
robert_harley6 wrote:Perhaps I'm misinformed, but my understanding is that a 16-bit word length can encode 65,536 discrete quantization steps. A 20-bit word length can encode 1,048,576 discrete quantization steps. I don't understand how a 16-bit word can encode more than 65,536 quantization steps.
You're not misinformed; a 16-bit word can indeed encode no more than 65536 steps.
We're not talking about encoding - they haven't encoded at 16 bits since the early 1990's. The issue here is re-dithering...and the preservation of 20-bit level resolution on standard 16-bit CD.
Read Atkinson for starters: www.stereophile.com/asweseeit/56/index.html
At the time, Atkinson suspected that only the "upper midrange" could be preserved with extended resolution. But it turns out that the lower midrange could be, too... and more than that still......
Then e-mail Bob Katz or any other audiophile recording engineer and ask them what the db specs are. Or read the liner notes in your CD booklet...I have one right in front of me from Sony Classical. It states - clearly - that they were able to preserve 20-bit sound on the 16-bit CD (using Sony's Super-Bit Mapping). And this disc dates from 1992 !!
In the end, I'm referring to the *sound* on CD, not the discrete levels, bits, etc.
Did you people think I was making this up ????
JPH
jph-22 wrote:
In the end, I'm referring to the *sound* on CD, not the discrete levels, bits, etc.
Did you people think I was making this up ????
Making it up, no. But (with respect) falling for the marketing BS, yes.
The editor of Stereophile would not "BS" any marketing hype - it's all science.
I have slightly re-stated my "20-bit resolution" statement in my opening piece and in my last posting. I now say "20-bit level" resolution. My original wording may have lead some to believe that I said CDs have 20-bits on them - they don't. But audiophile discs since 1993 have contained 20-bit *level* resolution, or noise floor, on them. It's analogous to a 4-cylinder car - it can have 150 horsepower or 200.
This should clear things up. And help more people understand why CD sounds so stunning through the latest playback gear. Is SACD better ? Reviewers Tony Cordesman and Jimmy Hughes heard some of the best CD players this year and think that it is not.....
I've just compared CD playback at what could be argued is the state-of-the-art (Reference Recordings HDCD-encoded discs played on the Spectral SDR-4000 Pro CD player) with the 176.4kHz/24-bit version of the same recording. I've also compared the CD with high-res through the same outboard DAC (Berkeley Audio Design Alpha DAC).
The high-res playback is from Reference Recordings' HRx format, which encodes the original high-res bitstream as WAV files on a DVD-R. You load the WAV files to a computer for playback through a Lynx AES16 card an an outboard DAC.
There's absolutely no question that the high-res is vastly more detailed, spacious, resolving, and dynamic. It's like throwing a light switch.
I go back to my original thought experiment: Does information contained in the 20-bit word's LSB survive in the 16-bit redithered word? If not, then redithered 16-bit digital audio is not the same as 20-bit digital audio.
Glad to hear this...but your comparison is limited. Earlier this year, you said that CD playback through Esoteric's reference system (w/ atomic clock) was like "great vinyl". If I read you correct, the hi-rez files are even better than Esoteric - meaning that they're beyond the *best* LP. Is this correct ?
In other words, we need a player that does CD justice...and Meridian's 808.2 will do that. You said (again) earlier this year that the Meridian 808.2 made CD sound like high-resolution versions of the CD copy. And since we know that upsampling does not add any information or resolution, there must be more on CD than we thought.........
JPH
I said that before I heard high-resolution done right. The high-res I'm listening to now is vastly better than vinyl, and has many qualities of a microphone feed (I've heard many hundreds of hours of mic feeds). High-res is so much better than the best CD playback, in my view.
I guess time - and more demos - will tell....
Because if your statement is correct, then hi-rez (now) is "vastly better" than the best LP - and this would be an all-time first. Never before has a reviewer claimed this - esp. with LP still advancing - like Clearaudio 150k system.
I mean, you did hear DVD-A - through Meridian - and I find it hard to believe that that wasn't "done right". How is hi-rez different now ??
JPH
It's hard to compare what I heard several years ago through the Meridian with what I'm listening to today.
My high-res system, reviewed in detail in the January issue, is a fan-less, drive-less PC (64GB of solid-state storage) for silent operation onto which I've loaded Reference Recordings HRx files at 176.4kHz, 24-bit files. I know the data being output to the DAC is a bit-for-bit identical copy of the original master (the intact HDCD flag in the data is the indicator).
This system has several benefits, not the least of which is the spectacular Reference Recordings sound quality. At the time of the Meridian, I didn't hear any recordings that were particularly well recorded.
DAC quality has improved as well in the interim.
Incidentally, the next issue includes an interview with Keith Johnson in which he talks about how close CD comes to high-res when done right, and how he used the high-res files as a benchmark in designing the Spectral SDR-4000 CD player.
Robert - there must have been *some* DVD-A's done right. I mean you did say that hi-rez, through the Meridian 808, was beyond its CD sound - which was then state-of-the-art.
I agree that Dac quality has improved... hence CD's better sound.
Listening impressions can be like living in a bubble - what we hear *at that moment* may not tell the whole truth. You stated many times throughout the late 90's and early 2000's that SACD was "vastly better" than CD. But others did not hear it this way. Neither HP nor JV thought SACD was better until the EMM gear of 2003. JV said in the June 2001 issue that SACD was "not an across-the-board improvement". Then HP's "The Truth about SACD" from early 2002.
It also true that every time a reviewer states that the new formats sound better, it's when they're listening to a "mixed" format player. These machines, unfairly, give an edge to these new formats. A dedicated CD player was needed to hear its truer potential...but very few reviewers told us this....
JPH
Post new comment