Quantcast
Redbook CD: What would specs be if it were invented today? - Page 2 - Gearspace.com
The No.1 Website for Pro Audio
Redbook CD: What would specs be if it were invented today?
Old 24th January 2013
  #31
Motown legend
 
Bob Olhsson's Avatar
 
Verified Member
🎧 15 years
It's like this: Sony and Philips wanted to bring a digital disk to market as quickly as possible that could be manufactured in their existing vinyl plants. For that reason the CD was built almost entirely around existing parts and technology. This was what set the technical limitations. There actually was also some kind of a lower resolution quad format option that I don't remember the details of.
Old 24th January 2013
  #32
113568
Guest
I think the answer would be spotify. No enduser ownership, just renting. No physical media. All done in cloud. Whatever audio quality the user wants/can stream

Sent from my XOOM 2
Old 24th January 2013 | Show parent
  #33
Gear Addict
 
🎧 10 years
Quote:
Originally Posted by The_K_Man ➡️
I don't know if you caught on but I did politely ask what "EFM" stood for and have yet to get an answer. I googled it but it stands for a dozen different things.
Here's your answer. http://www.exp-math.uni-essen.de/~immink/pdf/rey.pdf
Old 24th January 2013 | Show parent
  #34
Gear Addict
 
🎧 10 years
Quote:
Originally Posted by The_K_Man ➡️
I don't know if you caught on but I did politely ask what "EFM" stood for and have yet to get an answer. I googled it but it stands for a dozen different things.
Eight- to Fourteen-bit Modulation means that, instead of storing PCM directly on a CD, every 8 bits of data is converted into a different 14-bit code, which the CD player converts back to 8 bits to retrieve PCM for playback. Now that sounds a bit silly because obviously that means storing more data, but the clever part is that the 14-bit codes are all chosen so that each pit or land on the disc is at least 3 "bits" long, so for a given minimum spec of laser the data can be packed 3x more tightly and still be read, meaning you can fit 8/14*3 =~1.7 times as much data in the same space.

Without it, storing PCM at 44.1kHz 16-bit + PQ + checksums etc., discs would be less than 50 minutes long (my figures may not be perfect, there are some details I don't remember precisely!)


Incidentally, the CD standard itself is not necessarily limited to 2 channels - one of the subcodes (my hazy memory is telling me Q, although that doesn't seem right!) allows a flag to be set to denote a quad recording. Good luck finding a player that supports it though! I don't know if a quad-channel CD was ever actually released, or even whether the channel/speaker layout was ever actually defined.
Old 24th January 2013 | Show parent
  #35
Lives for gear
 
wado1942's Avatar
 
🎧 10 years
Quote:
Originally Posted by Hane ➡️
I'm not kidding, even 16 bit convertors didn't exit in the late 70's.
Sure they did, just not as an off the shelf component.
Old 24th January 2013
  #36
Lives for gear
 
🎧 10 years
I never understood, why the 16bits were used linearly. That's the only thing i would improve on a CD. More resolution torwards 0, less resolution torwards FS. Maybe then even 15 or 14 bit would be enough to mask quantize noise at all levels.
44,1kHz is fine, maybesomething around 60kHz or 88,2kHz for a good conscience, now that memory is not an issue anymore. Why anyone would want 96 or even 192kHz is beyond me.
Old 24th January 2013 | Show parent
  #37
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by haberdasher ➡️
Eight- to Fourteen-bit Modulation means that, instead of storing PCM directly on a CD, every 8 bits of data is converted into a different 14-bit code, which the CD player converts back to 8 bits to retrieve PCM for playback. Now that sounds a bit silly because obviously that means storing more data, but the clever part is that the 14-bit codes are all chosen so that each pit or land on the disc is at least 3 "bits" long, so for a given minimum spec of laser the data can be packed 3x more tightly and still be read, meaning you can fit 8/14*3 =~1.7 times as much data in the same space.

Without it, storing PCM at 44.1kHz 16-bit + PQ + checksums etc., discs would be less than 50 minutes long (my figures may not be perfect, there are some details I don't remember precisely!)


Incidentally, the CD standard itself is not necessarily limited to 2 channels - one of the subcodes (my hazy memory is telling me Q, although that doesn't seem right!) allows a flag to be set to denote a quad recording. Good luck finding a player that supports it though! I don't know if a quad-channel CD was ever actually released, or even whether the channel/speaker layout was ever actually defined.
Thanks to you - and all! This helps me understand a lot more about CD's development.

I'll say it again: It wasn't until I was twenty when I learned what AM and FM stood for, along with ABC, CBS, NBC, or CIA and FBI.

When "LOL" and SMH came along, I literally had a nervous breakdown! Point blank: I SUCK at abbreviations and acronyms!

So think about who might be browsing these threads, and adjust accordingly. I can learn - just not at the same pace as most of you might.
Old 24th January 2013 | Show parent
  #38
Lives for gear
 
wado1942's Avatar
 
🎧 10 years
Quote:
Originally Posted by evosilica ➡️
Why anyone would want 96 or even 192kHz is beyond me.
As I said previously, going from no attenuation to 96dB+ attenuation between 20KHz and 22KHz is no easy task. Doing the job right requires very complex analogue AND digital FIR filters and nobody has made a perfect anti-aliasing filter network yet, not in the analogue or digital domain... what we have now is simply "good enough for most cases". Even a so-called "ideal" anti-aliasing filter introduces a lot of issues in the audible band like preringing and increased quantization distortion. 96KHz would allow you to relax your filters so less damage is done within the audible band. You also don't have to rely nearly as much on reconstruction filters upon playback. Unfortunately, nobody designs AD and DACs with different filters for 96KHz vs the old standards. Even at that, almost all the junk created by the ADC stage is pushed outside the audible range. The same would be true for 64KHz using oversampling converters, but back in the day, even 64KHz wasn't enough and one early hard disk recording systems was 16-bit 100KHz because of this. BTW, by NASA standards, 200KHz was the suggested minimum to minimize sub-Nyquist distortion.
Old 24th January 2013 | Show parent
  #39
Gear Addict
 
🎧 10 years
Quote:
Originally Posted by wado1942 ➡️
The problem there was that channel separation with the matrix quad systems was at best 4.5dB; almost mono. Who wants to pay double for something that's inferior to 2-channel stereo? It was yet another deal where they pushed to get a product to the masses before it was ready. Later, they came out with logic steering, but you really can't have anything happening in adjacent channels at a time. Dolby's notion of L/C/R/S makes more sense, but that's not standard quad anymore and still doesn't allow for full mixes.
Plus too many incompatible formats using a poor choice of storage medium for a multichannel player, the vinyl record!



Quote:
Originally Posted by wado1942 ➡️

The best digital machines available off the shelf at the time was Mitsubishi's with 16-bit 50.4KHz. The worst machines at the time were 44.1KHz, 14-bit, but 3M's peculiar system was considered the most sonically neutral. It was 12-bit, 50KHz resolution with an extra 4-bit gain stage. So, it's 12-bit resolution, but 16-bit dynamic range because at lower levels, the gain stages would change to keep the signal quantization bits active. Philips's Videodisc, later known as Laserdisc, was the basis for the CD and it used a very good FM analogue carrier system, similar to VHS but without the obnoxious buzz that breathes with the dynamics of the material. The decision to go digital was a combination of insisting on a smaller size (going digital meant about half the area was needed) and Sony's vision of digital being the future.

Don't quote me on this, but I think THE first commercially viable digital multitrack tape recorder in 1972 used 13-bit 47.25kHz resolution. Really, Sony's 44.1K sampling was the worst that existed outside of prototyping experiments.
From what I've read, Denon made the first digital recordings in early 70's.


Quote:
Sure they did, just not as an off the shelf component.
I'd forgotten about the Soundstream digital recorders.
Old 24th January 2013
  #40
Lives for gear
 
ksandvik's Avatar
 
🎧 10 years
If I remember correctly, the first Phillips CD players were actually 14-bit with oversampling circuits.
Old 25th January 2013 | Show parent
  #41
Lives for gear
 
wado1942's Avatar
 
🎧 10 years
I don't think they started using oversampling till a year or two later, but I could be wrong on that. They still had direct converters, so lots of non-linearities.

Denon made the digital 8-track in 1972 that was technically a production unit. The first digital tape recorders were made in the 60s but only as prototypes.
Old 25th January 2013 | Show parent
  #42
Gear Nut
 
bandpass's Avatar
 
🎧 10 years
Quote:
Originally Posted by wado1942 ➡️
Even a so-called "ideal" anti-aliasing filter introduces a lot of issues in the audible band like preringing and increased quantization distortion.
Any ringing is at the filter's cutoff frequency, either 21 or 22 kHz (depending on whether you allow ultra-sonic aliasing)—not in the audible band. And if it's an analogue filter (or the analogue stage of the filter) then there's no quantization distortion; if it's digital, then the quantization distortion can be as low as you like.
Old 25th January 2013
  #43
Motown legend
 
Bob Olhsson's Avatar
 
Verified Member
🎧 15 years
The first Philips CD player used 4x oversampled 14 bit converters and most of us thought it sounded better than Sony's 16 bit non-oversampled player.
Old 25th January 2013 | Show parent
  #44
Lives for gear
 
wado1942's Avatar
 
🎧 10 years
Quote:
Originally Posted by bandpass ➡️
Any ringing is at the filter's cutoff frequency, either 21 or 22 kHz (depending on whether you allow ultra-sonic aliasing)—not in the audible band.
True, but some will argue that it still softens the sound of transients.


Quote:
if it's digital, then the quantization distortion can be as low as you like.
Correct me if I'm wrong, but isn't the DSP built into converters 24-bit? That can potentially limit the precision. On top of that, the filters used in converters aren't generally the best. There's a few that are pretty good, but still not perfect and you still really need analogue AND digital filters to get ideal performance.



Quote:
The first Philips CD player used 4x oversampled 14 bit converters and most of us thought it sounded better than Sony's 16 bit non-oversampled player.
Thanks for the clarification. I remember reading about Sony's being released first, then Phillips came out with their oversampling version. Wasn't it the late 80s before Sony did oversampling ANYTHING?
Old 25th January 2013 | Show parent
  #45
Lives for gear
 
dcollins's Avatar
 
Verified Member
🎧 15 years
Quote:
Originally Posted by ksandvik ➡️
If I remember correctly, the first Phillips CD players were actually 14-bit with oversampling circuits.
Yes. Four times oversampling.

Shannon, Beethoven, and the Compact Disc


DC
Old 25th January 2013 | Show parent
  #46
Lives for gear
 
dcollins's Avatar
 
Verified Member
🎧 15 years
Quote:
Originally Posted by wado1942 ➡️
True, but some will argue that it still softens the sound of transients.
You are probably thinking about pre-ringing at audible frequencies like when used in EQ.


DC
Old 25th January 2013
  #47
Gear Guru
There would never even be a CD format. Today, it would be MP3 bit reduced audio. We never would have got the higher resolution of 16 bit Red Book, much less SACD, it would be skipped for lower quality delivery systems, the one's most of the masses are perfectly happy with.

You got the audio you deserve, just like guvment.
Old 25th January 2013 | Show parent
  #48
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by dcollins ➡️
Yes. Four times oversampling.

Shannon, Beethoven, and the Compact Disc


DC
LOL!

I remember as a teenager hitting the electronics section of Macy's(when they had one) and saw "4X Oversampling" on the case of a consumer CD player. Then "8X" came along!

Was it just a marketing thing to put that on the components?
Old 25th January 2013 | Show parent
  #49
Lives for gear
 
wado1942's Avatar
 
🎧 10 years
Quote:
Originally Posted by The_K_Man ➡️
Was it just a marketing thing to put that on the components?
I'm sure of it, just like my first CD player advertised itself as having a 1-bit DAC. Oversampling really does solve a lot of problems, though. A lot of advertising gimmicks are just about taking the norm and coming up with a new name for it.
Old 25th January 2013
  #50
Lives for gear
 
ksandvik's Avatar
 
🎧 10 years
I still remember the a-ha experience using one of the first Phillips CD players such as truly hearing a concert pianist stomping on a pedal hearing the pedal mechanical sounds while performing. And the second non-a-ha when I saw all the companies re-releasing vinyl as CDs and forcing me to pay twice for the same product.
Old 25th January 2013 | Show parent
  #51
Lives for gear
 
dcollins's Avatar
 
Verified Member
🎧 15 years
Quote:
Originally Posted by The_K_Man ➡️
LOL!

I remember as a teenager hitting the electronics section of Macy's(when they had one) and saw "4X Oversampling" on the case of a consumer CD player. Then "8X" came along!

Was it just a marketing thing to put that on the components?
No, nothing to do with marketing. It was about conversion quality.
Old 26th January 2013 | Show parent
  #52
Lives for gear
 
Adam Dempsey's Avatar
 
Verified Member
🎧 10 years
Quote:
Originally Posted by The_K_Man ➡️
The reason I ask this question is because (1): I seem to get the impression that back when CD was new a lot of people, and not just audiophiles, seemed to have the same reaction to them when compared to LPs as many folks have today regarding MP3 when compared to CD! lol
The sonic issues were as much to do with A-D and D-A converter quality of the time (including stock PCM 1630 A-D's) and the release of CD's from digitized 2nd generation EQ'd-for-vinyl sources, as the format itself.
Old 26th January 2013 | Show parent
  #53
Gear Nut
 
bandpass's Avatar
 
🎧 10 years
Quote:
Originally Posted by wado1942 ➡️
Correct me if I'm wrong, but isn't the DSP built into converters 24-bit? That can potentially limit the precision. On top of that, the filters used in converters aren't generally the best. There's a few that are pretty good, but still not perfect and you still really need analogue AND digital filters to get ideal performance.
I'm not saying that there aren't some ropey ADCs out there, but 24-bit DSP is sufficient to maintain 16 noise-free bits: there's a square law, so the 8 additional bits will fill with noise (on average) after 256^2 = 65k arithmetic operations, so a FIR of a few thousand taps (or its equivalent using other techniques) is no problem.

Last edited by bandpass; 26th January 2013 at 11:23 AM.. Reason: typo
Old 26th January 2013 | Show parent
  #54
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by Adam Dempsey ➡️
The sonic issues were as much to do with A-D and D-A converter quality of the time (including stock PCM 1630 A-D's) and the release of CD's from digitized 2nd generation EQ'd-for-vinyl sources, as the format itself.
So CD was another "underutilized Porsche" back then, but for reasons dealing with converters.

I would still like to know what "oversampling" did, and why we don't see "4x", "8x", and 16x oversampling on the cases of CD players anymore.
Old 26th January 2013
  #55
Motown legend
 
Bob Olhsson's Avatar
 
Verified Member
🎧 15 years
Where do you see CD players any more?
Old 26th January 2013 | Show parent
  #56
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by Bob Olhsson ➡️
Where do you see CD players any more?
Electronic big boxes all keep at least one single player on the floor along with a carousel.

Real hi-fi stores have a selection of them. Just gotta look.
Old 26th January 2013 | Show parent
  #57
Lives for gear
 
dcollins's Avatar
 
Verified Member
🎧 15 years
Quote:
Originally Posted by The_K_Man ➡️
I would still like to know what "oversampling" did, and why we don't see "4x", "8x", and 16x oversampling on the cases of CD players anymore.
Principles of Digital Audio, Sixth Edition (Digital Video/Audio): Ken Pohlmann: 9780071663465: Amazon.com: Books
Old 26th January 2013 | Show parent
  #58
Lives for gear
 
Adam Dempsey's Avatar
 
Verified Member
🎧 10 years
Quote:
Originally Posted by Bob Olhsson ➡️
Where do you see CD players any more?
Hidden inside DVD and Blu-ray players.
Old 27th January 2013 | Show parent
  #59
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by haberdasher ➡️
Eight- to Fourteen-bit Modulation means that, instead of storing PCM directly on a CD, every 8 bits of data is converted into a different 14-bit code, which the CD player converts back to 8 bits to retrieve PCM for playback. Now that sounds a bit silly because obviously that means storing more data, but the clever part is that the 14-bit codes are all chosen so that each pit or land on the disc is at least 3 "bits" long, so for a given minimum spec of laser the data can be packed 3x more tightly and still be read, meaning you can fit 8/14*3 =~1.7 times as much data in the same space.

Without it, storing PCM at 44.1kHz 16-bit + PQ + checksums etc., discs would be less than 50 minutes long (my figures may not be perfect, there are some details I don't remember precisely!).
No.

CD is coded so that transition from "pit" (which actually is a "ridge" from the side laser sees it) to "land" is a 1 and everything between the start and the end of a pit or land is a string of zeros. Thus it is the change from reflection (land) and no reflection (pit/ridge) or vice versa which is 1 and between each one there are a string of zeros, minimum of 2 and maximum of 10 between each 1. To convert 8 bit words into words which fulfill this criteria of 2-10 zeros between each one at least 14 bits are needed. This conversion happens in CD player hardware using a simple lookup table.

In addition there are 2 connection bits between each 14 bit EFM word to ensure that there are no less than 2 zeros between words in case they end and start with 1, and also one extra bit which is used to flip the phase of the signal as needed to avoid DC buildup (not audio, but data signal). So the coding is quite wasteful, as 17 bits are needed for each 8 bit data word. In addition there is error correction redundancy in the signal and data blocks are shuffled around for interleaving.
Old 28th January 2013 | Show parent
  #60
Gear Addict
 
🎧 10 years
Quote:
Originally Posted by Petrus ➡️
No.
Hey that's a bit much, I said there were some details I didn't remember!

Quote:
CD is coded so that transition from "pit" (which actually is a "ridge" from the side laser sees it) to "land" is a 1 and everything between the start and the end of a pit or land is a string of zeros. Thus it is the change from reflection (land) and no reflection (pit/ridge) or vice versa which is 1 and between each one there are a string of zeros, minimum of 2 and maximum of 10 between each 1. To convert 8 bit words into words which fulfill this criteria of 2-10 zeros between each one at least 14 bits are needed. This conversion happens in CD player hardware using a simple lookup table.
Yep, so the string 100, for example generates a pit or land the equivalent of 3 bits in length. I omitted the maximum because I couldn't remember the figure and because it didn't matter too directly to my point.

Quote:
In addition there are 2 connection bits between each 14 bit EFM word to ensure that there are no less than 2 zeros between words in case they end and start with 1, and also one extra bit which is used to flip the phase of the signal as needed to avoid DC buildup (not audio, but data signal). So the coding is quite wasteful, as 17 bits are needed for each 8 bit data word.
And there's the, er, bit I forgot, thanks! So in terms of minimum pit/land length EFM allows the amount of data stored to be 8/17*3 =~ 1.4 times as much, not 1.7.

Quote:
In addition there is error correction redundancy in the signal and data blocks are shuffled around for interleaving.
Certainly.
📝 Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 19 views: 2213
Avatar for Gretschman
Gretschman 1st November 2007
replies: 1 views: 1644
Avatar for neptunesorbit
neptunesorbit 21st November 2013
Topic:
Post Reply

Welcome to the Gearspace Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…

Forum Jump
Forum Jump