Griffinator, et al who participated in the "Harsh CD" Thread...

[Mallette]

>OK, about your equipment. The reason I kept asking about it? It's real simple. When the ADC in your recording rig is located inside your computer (as the WamiRack is) it is subjected to a barrage of EMI from the CPU and other extraneous internal components.

The Wami DAC is in the rack portion some 15 feet from the computer. The card portion is simply an I/O to the bus. While some cheap cards are probably more or less subject to this problem (actually, I've heard the effects with some), my completely internal card I use for recording as well is a Card Deluxe, which is well shielded and quite immune. The CDeluxe people were aware and concerned about this, since part of it's design is for scientific purposes.

>This has a huge impact on the resulting capture. I submit to you that the WamiRack at 192Ks/S is certainly going to be much better not because it captures 96Khz frequencies, but rather because at that samplerate it can self-correct the jitter errors, where at 88.2 and certainly at 44.1 it cannot.

This is more towards the point of the discussion, and I'll grant you this as a possibility. Most errors occur WAY into the ultrasonic at 24/96 and above. However, this in itself would be adequate to suggest that the Redbook spec is fatally flawed by the need at the time of it's conception to limit data to the available medium. DVD-A is similarly limited by spec for the same reason. With blue laser technology on the horizon that will effectively (for audio purposes) eliminate the storage issue, it is time to set a standard that does not take into account media capacity.

As for jitter, I've not been entirely convinced it's an audible issue. Not saying it isn't but just saying it is not proven to my satisfaction.

>Prove to yourself and to me that you have actually captured ultrasonic information. Until you do that, you are "going on faith" that ultrasonic information actually exists in your recordings.

While I've dwelt on this issue of the presence or lack of ultrasonics in recordings, it's mainly because my muse had settled on it lately as a possibility in the issue at hand. Though not settled, I am less obsessed with it than 24 hours ago and considering other things, like sample rate.

It has been debated since the early days of Redbook whether 44khz (ultrasonics aside) was a small enough slice. That is, that somehow the human brain could sense a slice size that large as "grain" just like in a photo.

Point is, we've been through some very fascinating discussion here and the mystery remains. There is no question that 24/192 can sound downright REAL, 24/96 very realistic, and below that just adequate.

Dave

[fini]

----------------

On 8/4/2003 7:24:35 AM Marvel wrote:

Did I tell you of a friend of mine who works at the Edison Historical site in New Jersey? They have recorded on Edison's original equipment, with current talent.They are then transfering to CD.

----------------

A favorite example of this is "I Can Hear You," by They Might Be Giants, from their album, "Factory Showroom." It's also available on their 20-year retrospective double-album, "Dial-A-Song" (highly recommended). Here's the mpg.

fini

[Mallette]

Interesting. The brick cieling on that was VERY low indeed, I presume due to extreme compression.

The detail within the available area is striking, however.

Let me clarify the first line. My computer room speakers are quite good, and there was a terrific distortion edge and limit beyond which there was nothing. It did not sound like a limitation of the cylinder, but of the mp3.

Very relevant to the discussion, since I can't really tell which it is...though I suspect the digital end. Analog distortion or overdrive is rarely that unpleasant. I kept wanting to STOP it, though I listened to the whole thing.

Dave

[Marvel]

Dave,

That TMBG file was encoded into a flash file -- could very well be why it sounds the way it does. Mucho compression!

This might explain fini's sense of humour as well. Next we'll be listening to the Firesign Theater.

Marvel

[fini]

Dave,

I didn't listen to the mp3, but the sound quality of that cut is awful on the cd, but that's the point of the song. Are you familiar with They Might Be Giants? They do songs about James K. Polk, mammals, a worm that isn't a real doctor, singing like a girl, and (perhaps their most famous tune) one about a night-light. There's even a film out about them! Check it out! Coming to a theater near you! If you didn't catch the lyrics:

I CAN HEAR YOU

I can hear you

Just barely hear you

I can just barely hear you

This is a warning,

step away from the car

This car is protected by Viper

Guess where I am

I'm calling from the plane

I'll call you when I get there

You won't hear a buzz

But I'm buzzing you in

I'm buzzing you in

What's your order?

I can super-size that

Please bring your car around

I can hear you

Just barely hear you

I can just barely hear you

[Mallette]

Actually, the lyrics were pretty clear. Can't imagine why they'd do something like that to make it unlistenable. No early recordings in good condition and played back properly are anywhere near that painful.

Dave

[fini]

LOL!

I particularly like these lines:

What's your order?

I can super-size that

Please bring your car around

When have you ever hear these clearly?

fini

[Griffinator]

The crux of my position is simply that the quality of the filter systems in modern ADCs and DACs are what determine the quality of their output at any sample rate. It just so happens that it's a lot easier to manufacture a transparent HP filter when the slope can start at 20K and very gently roll off to 96K or even 192K. It's a lot more difficult (and more expensive) to produce a similarly transparent filter when the slope must start at 20K and roll off to 24K. So the lower price point manufacturers were thrilled to jump on the 'high rez' bandwagon and hand out whatever pseudo-science they could get their hands on to convince people that anything other than poor quality components necessitated these high bandwidth recording formats.

I never said 16/44.1 was the end all - but I feel strongly that 24/48 is all that is necessary, provided the converters are properly built (and there are many out there that meet this standard).

My AI3's are not in this category, but they are all I can afford at this point. My next step will be Genex DSD/PCM 8 channel ADCs - but that's a big step - $4000 per box plus interface card plus software upgrade (Pyramix studio) - but the end result is my studio being fully DVD-A and SACD surround capable - a worthy investment, IMO.

[Mallette]

Cool enough. I've got questions, not answers.

But why spend money trying to make up for the issues involved with the lower sample rates when higher rates are only limited by storage capacity? Storage gets cheaper constantly, and blue laser technology with 100 gig DVD size discs at 50 cents each will soon pretty much eliminate any need to worry with file size.

I understand the current investment in the 70's technology, but it is inherently unfixable in the sense of making it work for the 100.00 or so most want to spend on a player, at least in the foreseeable future. Even audiophiles with deep pockets have issues with 15,000.00 Mark Levinson DACS that seem to be required to make a CD sound anywhere near as good as my 2000.00 investment.

Dave

[Griffinator]

Ahhh - see, you're thinking on the consumer side, I'm thinking on the producer side.

It's not just the storage that makes me shy away from higher sampling rates - it's the clock cycles and disk access speeds those higher rates demand from my system. Very hard to be creative behind the desk when you're restricted by whether your CPU will hold out or whether a drive in your RAID will crash...

[Mallette]

Now we're coming together!

Those are issues. I had a few of 1/30 second or so drops in my 24/192 location recording, due, I believe, to thermal recalibration or simply pure data overload to the drive. Luckily, there were two shows and more than enough material to make a disc.

These are temporary issues. I am going serial ATA next time out. I think they are more than fast enough to handle the load. Further, until there is a releasable 24/192 format, I am going to drop to 24/96 for 4 or more channels. I can hear the difference (or so I believe), but it is pointless to tax the equipment with a data rate and format you can't release and nobody can play back.

OTOH, I will probably stay with 24/176.4 for stereo since I, at least, will be able to enjoy the high res and the reduction to CD or whatever will be nice and even with truncation and no dither needed.

Dave

[Marvel]

I see some issues here. At this point, CDs are still 44.1k, yet you would want to use 48k? Or, you think that is all you really need?

Griff is stuck, if you will permit my use of the phrase, with an interface that will only do up to 48k, no matter what the rest of his system will do. DATs, ADATS, video recorders have that as their max sample rate. Dave would still like to figure out if ultrasonic frequencies have any impact on recorded audio, so the LP filter idea is reasonable, but there must be more to it than the quality of the filters. I have a friend whose last two albums were done with Pro Tools, which version I don't know. They are mostly acoustic, and are simply stunning. I don't agree with the whole Pro Tools argument, but it is hard to argue with how good it can sound, even on CD. A question I would have is would it sound even better to hear it mixed at 192K?

Are these the main questions right now (since Griff brought up 48K issue)?

1. Is it just the quality of the filters on the ADC and DAC?

2. Do we really require more than a 48K sample rate.

3. Can we test for whether or not ultrasonics affect the audio we hear?

Marvel

[Mallette]

I've backed off the ultrasonics a bit. Still not settled about it, and certain there is a fundamental flaw in 16/44.1, but I suspect it to be trickier than some basic measurable issue. In my case, while I care, it is not my primary aim to solve this mystery. For my own purposes, it is conclusively proven:

24/192=As real as the playback system will allow

24/96=Just short of the very finest LP's

16/44.1=won't fool a cat on acoustic instruments

Further, I've determined that Redbook is fatally flawed in laying off too much correction and reconstruction to the hardware. In practical fact: You shouldn't need a 5,000.00 CD player if the medium is any good.

Now, this is beginning to sound like an attack on CD. It isn't. I'll still be making them because they are a fact of life. However, you can't argue with ears.

As to Marvel's question about whether "it" (whatever "it" was) might sound better mixed at 192k, I don't know. If no software processes (eq, filters, etc.) are used, I doubt it. If such software efx are applied, probably. It is generally accepted that the greater redundancy of data, the less the audible impact on the result.

I don't use such things, so I cannot verify this.

Best regards,

Dave

[Griffinator]

----------------

On 8/6/2003 10:14:40 AM Marvel wrote:

I see some issues here. At this point, CDs are still 44.1k, yet you would want to use 48k? Or, you think that is all you really need?

Griff is stuck, if you will permit my use of the phrase, with an interface that will only do up to 48k, no matter what the rest of his system will do. DATs, ADATS, video recorders have that as their max sample rate. Dave would still like to figure out if ultrasonic frequencies have any impact on recorded audio, so the LP filter idea is reasonable, but there must be more to it than the quality of the filters. I have a friend whose last two albums were done with Pro Tools, which version I don't know. They are mostly acoustic, and are simply stunning. I don't agree with the whole Pro Tools argument, but it is hard to argue with how good it can sound, even on CD. A question I would have is would it sound even better to hear it mixed at 192K?

Are these the main questions right now (since Griff brought up 48K issue)?

1. Is it just the quality of the filters on the ADC and DAC?

2. Do we really require more than a 48K sample rate.

3. Can we test for whether or not ultrasonics affect the audio we hear?

Marvel

----------------

First of all, I'm not "stuck" with the AI3's I use. I also use dBx tube channel strips that will support 192Khz - but I don't care to use the digital outs on those strips. I route the dBx's straight into the Alesis.

We can, and people have repeatedly, tested the perceptibility of ultrasonic information. It's a simple A/B test - record a source with known ultrasonic content (cymbals are an excellent choice) at 24/192 (or 24/96), use a lowpass filter set to 20Khz to roll off the high end on one version of the source, and blind-audition them on a system that can handle these ultrasonics (most modern receivers will generate ultrasonic content, albeit at a slightly lower dB response than 20-20K material).

And yes, there is a growing contingency of recording engineers, producers, and mastering engineers that are becoming convinced that higher bandwidth digital recording specifications are merely an out for manufacturers who choose not to implement properly designed comb filters, HP filters, and reconstruction filters in their ADC and DAC circuitry. That said, nearly all of us support 24/48 specifications as the minimum necessary to transparently capture and reproduce audio, and that the Redbook format is not a sufficient reproduction medium.

[Mallette]

Griff: This is neither a challenge nor necessarily even disagreeing.

I'd just like to point out that about every couple of decades a new standard is proposed that is the last word in audio reproduction. I personally find it very difficult to believe that the modest move to 24/48 is going to fool my cat. BTW, I mean that literally. She will hardly stay in the room with 24/96 or above crowd noise as she constantly looks towards various sources as they happen. She sleeps like a kitten with even loud CD's.

I suppose we could call this one "Mallett's Cat" with apologies to Schrodinger...;-)

While 100 gig DVD's are still 2 years or so from general circulation, I believe the specifications should be broadened to allow issuing in at least the highest resolution formats we now have.

As to the reliability of A/B tests, remember the one I mentioned from 1912 or so. I don't think they are reliable. If I were going to construct something that MIGHT be valid, it would involve putting listening equipment in a number of houses and doing such a test over months, if not a year. It took me a decade to realize that CD's were not feeding my soul.

A person can starve to death so gradually they are not even aware.

Dave

[Marvel]

Almost went right past it, but a hearty welcome to Lynn Olson. Is that an Ariel or ME2 in your avatar? I would love to hear the amps described on the Nutshell website.

Griff,

I'm not trying to argue with you. I just understood that the AI3 analogue inputs will only give you up to 48K out on the lightpipe. Am I wrong about that? If you only use that for your input, then you are stuck, no matter what the dbx strips will do. It wasn't meant as a put down.

I see you and Dave as doing two different kinds of recording, and Dave mentioned a problem with the higher samplerate. TOO MUCH data. If you are runnign a studio, you are probably going to have way more tracks than he does, and it will get very hard to stream that much data off the drives if done at a higher sample rate.

The idea is to learn and share information, not to diss anyone.

Marvel

[LynnOlson]

Hi Marvel and all the other great Klipsch posters, glad to be here!

At the 2001 VSAC show here in Silverdale, Stan Ricker, Paul Stubblebine, Ed Meitner, and Mike Pappas put on a hi-rez demonstration using professional-grade Meitner Design convertors, using material they had recorded themselves with analog 30IPS, 192/24, 176/24, 96/24, 44.1/16, and Sony DSD. Despite playing to a fairly large crowd in a hotel conference room - hardly an ideal demo environment - I could hear pretty obvious differences from my listening position about five rows from the front. At the highest level - analog tape with custom electronics, DSD, and 192/24 - the differences were pretty small and could be put down to personal taste. Different members of audience had different preferences. The three top choices sounded impressively realistic, with a strong "hall sense", very natural tone-color, and startling dynamics.

The step-down in quality to 96/24 was heard by most members of the audience, and was pretty obvious to folks in the front rows. I heard a slight but noticeable coarsening of sonics, some of tone color, and a more closed-in impression of hall space. Acceptable overall, but in truth not a completely transparent medium.

The further reduction to 44.1/16 Red Book, even with pro-grade Apogee converters handling the dithering and resampling duties, with disliked by everyone, and sounded well onto its way to low-fi MP3 or AC3 compressed audio. Hall sense was nearly absent, an obvious metallic coloration affected the string tone, and some instruments like xylophones were grossly altered in tone color. Compared to the mike-feed quality sound we'd been listening to, this was not much better than mid-fi, and a devastating comment on just what poor sound we've been accepting for two decades.

My own thoughts about the pro-quality demo (and a big thanks to "Doc" Johnson for the recordings) was that consumer-quality SACD is far below pro grade, and that 30IPS two-track analog tape, 192/24, and DSD are surprisingly close in quality. Each had subtle defects, but they were close to the limit of perception, and only apparent on direct comparison with the same recording using a similar medium. 96/24 was still High Fidelity in the accepted sense of the term, did justice to the recordings, and was pretty close to the Top Three in quality. Red Book fell well short of the mark if musical realism, or even musical values, were important to the listener. 44.1/16 is better than MP3 or AC3, but that ain't saying much.

[LynnOlson]

Sorry to post again, but I didn't answer Marvel's question: yes, that's an Ariel you see on the avatar. The photo was taken at the disastrous Portland Oregon Triode Society demo, with the poor Ariels driven by a Krell KSA-250. Man, do they suck with Krells!!!

Later at the demo, they were connected to VAC PA90's and sounded wonderful - but by then most of the audience had drifted off and were looking at other items in other parts of the building. When I tell people the Ariels (or ME2's) don't work with most transistor amps I'm not kidding.

As for the stuff on the Triode section of the Web-page, I'd say there's an excellent chance that Gary Dahl will bring his Aurora's to the 2003 VSAC and play them on his Chorus II's. He's probaby listening to them right now!

[Deang]

This has got to be one of the most depressing threads I've ever read.

I would imagine the recordings used had little or no compression? Is there any chance in hell that the noted, perceived differences would be minimized in commercial recordings for the general consumption of us drones?

[Marvel]

Dean,

I did some studio work for friends of mine in the early '70s. The studio did 24 tracks at 30ips. You go through a lot of tape that way! There is incredible headroom at 30ips, as well as at 24/192. Compression probably wouldn't enter into it much. Still, getting that onto a Redbook CD means something has to go.

Lynn,

Thanks for the post. I'm probably living in ignorant bliss for having a lot of commercial CDs that I think sound very good. We always do until we hear something better I guess.

Marvel