Why 24/192 is a bad idea?

[bdiament]

Hi Amir,

What is "word length clipping"? I've not heard the term before.

When I record (i.e. the first digital iteration from an analog signal), I make sure the max peak doesn't exceed -6 dBFS. In my experience, this is where A-D converters exhibit their lowest distortion. Final levels are adjusted during mastering.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

[Phelonious Ponk]

I'm pretty sure I don't fully understand this rather technical discussion, but I do get the part about building excess headroom into files, and I'd just like to say that, as a consumer of digital media, if 16 bits is not quite enough to drive all the distortions below the noise floor at listening levels of 120 dB (peak) and we need to go to 17, even 20, I'm fine with that. I just don't want to sit around waiting for something to download (or pay a premium for a perceived advantage...and you know that's what will happen) for anything that doesn't deliver a tangible benefit. If the current standard is not quite up to par, bring it up to par. Don't over engineer for the sake of it, because it will creat unnecessarily large files and set the marketing guys off on the trail of deeper BS and higher prices.

Of course this is all hypothetical. The general public doesn't hear this stuff. The marketing of higher and higher rates will be aimed at the specialty audience, Audiophiles, and as always, the differences will be astounding.

Tim

 

[bdiament]

Hi Tim,

I don't know anyone outside of one or two record producers who would endure 120 dB levels in their listening room - even if they had gear that would not self-destruct in the attempt to reach those levels.

As far as the general public, I've brought more than a few "civilians" into the studio and played them 24/192 files vs. 16/44 files. They've all heard differences here. How important those differences are, vary with each individual.

"Tangible benefit" is definitely in the ear of the behearer. Listening to my own recordings or making comparisons of Keith Johnson's wonderful recordings in both CD versions (among the best sounding CDs I've heard) and the 24/176 versions, I personally hear quite a few "tangible benefits". From my own work - where I have the benefit of standing at the position of the microphone array during the recording sessions *and* of listening to the direct mic feed prior to recording - I find it quite easy to hear: The high res version is something I have not been able to distinguish from the mic feed; the CD version, good as it might be is not remotely close (and because the 24/192 version sounds like the mic feed to me, the 24/96 version, good as it might be, sounds closer to the CD version than to the 24/192).

But I understand that's just me and the gear I'm recording and listening with. As I've said many times before, a lot of the converters I've heard, which have a "192k" on their spec sheet, are not delivering on the potential I've heard realized by other converters. And of course, there is the variation from listener to listener. In my experience, different folks have different sensitivities to different aspects of sound. For example, some are sensitive to dynamic compression while others aren't. Some hear those quite tangible benefits with 4x rates (176.4k and 192k) while others don't. Those that find no benefit can certainly save on the cost of recordings and on hard drive space.

By the way, in view of how few converters in my experience can actually deliver fully on the 4x rates, I find specs citing even higher rates (e.g. 384k and 32-bits) to be more than a little bit silly. Just my perspective of course. (I note, perhaps coincidentally, perhaps not, that the makers of converters that do the job at 4x rates don't make such claims.)

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

[Bruce B]

Like in this HD-Tracks high-res release of Norah Jones, which also illustrates the point about falling energy levels the higher the frequency, until the 22.5kHz point, when the tweedling begins.
http://www.computeraudiophile.com/a...27-norah-jones-come-away-me-hdtracks-four.jpg

This is an exception to the rule. There are very few that I've seen drop off this quickly. Most of the time it's very linear.

 

[Phelonious Ponk]

Hi Tim,

I don't know anyone outside of one or two record producers who would endure 120 dB levels in their listening room - even if they had gear that would not self-destruct in the attempt to reach those levels.

As far as the general public, I've brought more than a few "civilians" into the studio and played them 24/192 files vs. 16/44 files. They've all heard differences here. How important those differences are, vary with each individual.

"Tangible benefit" is definitely in the ear of the behearer. Listening to my own recordings or making comparisons of Keith Johnson's wonderful recordings in both CD versions (among the best sounding CDs I've heard) and the 24/176 versions, I personally hear quite a few "tangible benefits". From my own work - where I have the benefit of standing at the position of the microphone array during the recording sessions *and* of listening to the direct mic feed prior to recording - I find it quite easy to hear: The high res version is something I have not been able to distinguish from the mic feed; the CD version, good as it might be is not remotely close (and because the 24/192 version sounds like the mic feed to me, the 24/96 version, good as it might be, sounds closer to the CD version than to the 24/192).

But I understand that's just me and the gear I'm recording and listening with. As I've said many times before, a lot of the converters I've heard, which have a "192k" on their spec sheet, are not delivering on the potential I've heard realized by other converters. And of course, there is the variation from listener to listener. In my experience, different folks have different sensitivities to different aspects of sound. For example, some are sensitive to dynamic compression while others aren't. Some hear those quite tangible benefits with 4x rates (176.4k and 192k) while others don't. Those that find no benefit can certainly save on the cost of recordings and on hard drive space.

By the way, in view of how few converters in my experience can actually deliver fully on the 4x rates, I find specs citing even higher rates (e.g. 384k and 32-bits) to be more than a little bit silly. Just my perspective of course. (I note, perhaps coincidentally, perhaps not, that the makers of converters that do the job at 4x rates don't make such claims.)

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

Perhaps my silly bar is just set a bit lower than yours, Barry . Don't get me wrong, if there's a benefit to the higher rates, I think we should go for it, particularly in this new age that doesn't require custom media and players. But I know marketers, and I know Audiophiles, and I know that if the benefits of high resolution aren't clearly defined, resolution rates and the hyperbole that accompanies them will get ever sillier and many Audiophiles, as always, will hear every dime they invest, whether it is silly or not.

Tim

 

[bdiament]

Hi Tim,

I agree insomuch as when there is an opportunity for commerce, someone will take that opportunity. And also, it would seem, there is a customer for everything.

However, as a music and audio enthusiast, those realities don't impact the real pleasure I take in listening to well made recordings delivered at high resolution. The "clearly defined" those benefits are seems to vary from person to person, as we can clearly see in threads like this one. (This is why I created the Format Comparison page on the Soundkeeper site; so folks can download and listen to samples on their own systems and make their own determinations.)

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

[microstrip]

This is an exception to the rule. There are very few that I've seen drop off this quickly. Most of the time it's very linear.

Shostakovitch 5th Synphony, last movement - plenty of energy! A serious test for any system.

 

[DonH50]

A bit off topic but since it has been mentioned, some information:

In the old days of digital, it was thought resolution was maximized by ensuring the loudest sounds peak at 0 dBFS ("0 dB Full Scale"), the loudest that can be captured by a digital recording system.

With every ~6 dB (actually, more like 6.02 dB) representing one bit in the digital "word", lower level signals are captured with ~1 bit less resolution for every 6.02 dB below 0 dBFS. Translated to English, this means, with uncompressed audio, where the average level can be 20 dB or more below the peak level, the average level parts of the audio signal would be captured with >3 bits less resolution than the loudest peaks, assuming the level of the loudest peaks was 0 dBFS. With a 16-bit system, that means the average level sounds in a recording would be captured using the bottom 12 or 13 bits of resolution. Other sounds in an uncompressed recording may be well down in level from there and end up being captured by considerably fewer bits. Astute listeners will hear the coarsening of harmonic structure that results from the lower resolution, as well as the defocusing, if not complete obliteration of spatial information. (I talked about this in a little more detail in post #25 on page 3.)

They key I want to mention in this post is that most of my colleagues have come to avoid having the maximum peak at 0 dBFS and today, will leave anywhere from 0.3 dB to a dB (sometimes more) headroom. The current thinking is to see 0 dBFS as the equivalent of an overload and in fact, the meters on some A-D converters will now go "into the red" when the signal reaches 0 dB, rather than waiting until that level is exceeded.

It turns out a number of D-A converters, particularly cheaper ones, can "stick" when the signal gets to 0, producing artifacts, sometimes manifesting as low level clicks, sometimes worse. Moreover, it is possible for the peaks in the reconstructed analog signal to be higher than the values represented by the samples on either side of the peak (hence, they are called "intersample peaks"). Avoiding max peaks at 0 dBFS eliminates these problems.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

Nice post, Barry.

IME, most converter testing is done at -1 dBFS to avoid clipping. "Digital" clipping at the output of a DAC can be very harsh as most DACs have very large bandwidth. Again in my limited experience and through discussions, max (loud) average levels are typically set at 20 - 30 dB below full-scale during recording to avoid clipping on uncompressed music or live recordings. The loss in low-level detail and effectively higher noise/distortion due to the reduced number of bits used is one of the drivers for higher resolution (e.g. 24-bit) ADCs and DACs. Final mastering often compresses the dynamics.

The clicks are likely due to d.c. offsets induced by clipping, and in particular asymmetric clipping. It could also be due to glitches over-driving the output buffers; some opamps will latch or even invert their outputs when overdriven, leading to all kinds of nastiness.

I thought I had posted this someplace, but for reference if you start with a 16-bit DAC having 98 dB SNR, then give up 24 dB (4 bits) for headroom you are left with 12 bits or 74 dB SNR. If your lowest sounds are say 42 dB below that, you lose another 7 bits and your lowest signals are being recorded with only 5 effective bits (32 dB SNR), and of course lower-level components use even fewer bits (e.g. a soft piano or guitar note may sound harsh asit decays due to only a few bits being used). The quantization noise floor does not reduce as the signal is reduced. The usual counter argument is that those sounds are so quiet we don't really hear them anyway. However, one of the reasons dither was added to early converters was because the noise floor sounded harsh and harmonics were (are) added to low-level signals; dither helps spread the energy out (through decorrelation) and makes the noise floor more "analog" or "smoother" sounding.

 

[bdiament]

Hi Don,

Thank you.
I believe I mentioned in an earlier post (perhaps in a related thread on another forum) that during recording, I don't allow the max peak to exceed -6 dBFS.

In my experience - at least with all the A-D converters I've tried - A-D conversion exhibits lower distortion this way. Final levels are adjusted in mastering. (I don't use compression in my work, recording, mixing or mastering; I *like* dynamics. ;-})

Since I record at 24-bits, I'm not at all concerned with losing low level information. For me, it isn't a question of noise (at all) but rather the *resolution* of that low level information (where instrumental harmonics and spatial cues live), particularly in view of the fact that I leave performance dynamics fully intact.

As to those sounds being "so quiet we don't hear them anyway", I would beg to differ, especially with wide dynamic range recordings. (I fully understand that some folks may not hear them. But clearly, others do.) To my ears, at 16-bits, that information gets coarsened, when it isn't completely obliterated. At 24-bits, it is there to be heard. To borrow a visual term, which I believe still applies to audio, the results exhibit much better *focus*; it is easier to hear the fine nuances of instrumental and vocal sounds and easier to hear the space around the players. A 16-bit version of the same thing just sounds (to me) at best like the focus wasn't properly locking in and at worst like it wasn't even attended to.

This is why I don't argue with folks that don't hear the difference or with folks that argue that 24/192 is excessive or somehow worse than lower rates. I'm hearing the best recorded sound I've every experienced and someone's trying to tell me I'm wrong, that it is a "bad idea". (!?) As I've said elsewhere, it is like someone telling me there aren't any colors in a rainbow.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

P.S. That last paragraph reminds me of what the manager at a large NYC studio once told me. He said that women do not have orgasms. He told me he'd been with many, many women over the years and not once did he witness one having an orgasm. =8-0

 

[Ethan Winer]

The usual counter argument is that those sounds are so quiet we don't really hear them anyway.

Yes, that's certainly my usual counter argument.

Seriously, in order to actually hear the lower quality of having only 5 bits to capture very soft content, you must raise the volume beyond its usual setting. I don't listen that way, and none of my friends do either. That might be useful to assess noises to see where a buzz or whatever might be coming from. But it's not how we listen to music.

--Ethan

 

[rbbert]

Hi Tim,

I don't know anyone outside of one or two record producers who would endure 120 dB levels in their listening room - even if they had gear that would not self-destruct in the attempt to reach those levels.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

Barry, on your recordings where there are percussive peaks, how far above the average RMS volume are they? 20 dB? 30 dB? If the latter (and I suspect they are close), and your average listening volume is ~85 dB, then your peaks are 110-115 dB, right?

 

[bdiament]

Hi rbbert,

Barry, on your recordings where there are percussive peaks, how far above the average RMS volume are they? 20 dB? 30 dB? If the latter (and I suspect they are close), and your average listening volume is ~85 dB, then your peaks are 110-115 dB, right?

It varys from track to track. So far, some have been in the high 20s above the RMS.

You are correct that for an ~85 dB listening level, the peaks will then be at ~110 dB. (I think when I first read Tim's post, I was thinking *average* level, which some producers I've known will subject themselves to.)

Still, I say again (or did I say this in a related thread on another forum?), it isn't the noise floor I'm concerned with, it is the *resolution*. This is unrelated to playback level, though I do find wisdom in Peter Walker's words to the effect that every recording has one "correct" playback level.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

[rbbert]

...Still, I say again (or did I say this in a related thread on another forum?), it isn't the noise floor I'm concerned with, it is the *resolution*. This is unrelated to playback level, though I do find wisdom in Peter Walker's words to the effect that every recording has one "correct" playback level.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

But again, from earlier in this topic, the engineers are not accepting that bit-depth has any meaning except establishing the noise floor, so I'm restricting my arguments only to that, in order to support the concept that a minimum of 20 bits is necessary for music recording. I understand perfectly what you mean when you say "resolution", but that's not an argument you will win with some ostriches.

 

[Ethan Winer]

the engineers are not accepting that bit-depth has any meaning except establishing the noise floor

As soon as someone shows hard proof - not subjective prose - that anything more than the noise floor is affected, I promise you that every real EE type audio engineer will change their opinion. So whatcha got?

--Ethan

 

[bdiament]

Hi rbbert,

But again, from earlier in this topic, the engineers are not accepting that bit-depth has any meaning except establishing the noise floor, so I'm restricting my arguments only to that, in order to support the concept that a minimum of 20 bits is necessary for music recording. I understand perfectly what you mean when you say "resolution", but that's not an argument you will win with some ostriches.

I'm not arguing. ;-}
Neither are the designers of my favorite gear.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

[rbbert]

As soon as someone shows hard proof - not subjective prose - that anything more than the noise floor is affected, I promise you that every real EE type audio engineer will change their opinion. So whatcha got?

--Ethan

I've got the fact amply documented by both Amir and common sense (empirical observations which you seem to value highly) that you need a minimum of 120 dB dynamic range to transparently record that aspect (signal above noise) of music. That's quite a bit more than 100 dB (to say nothing of 80 dB). I'm consistently avoiding the "noise floor vs. increased resolution" issue.

 

[Bruce B]

I've got the fact amply documented by both Amir and common sense (empirical observations which you seem to value highly) that you need a minimum of 120 dB dynamic range to transparently record that aspect (signal above noise) of music. That's quite a bit more than 100 dB (to say nothing of 80 dB). I'm consistently avoiding the "noise floor vs. increased resolution" issue.

Especially if you archive vinyl with a DR >120dB - re: Keith O. Johnson

 

[DonH50]

You have vinyl LPs with >120 dB dynamic range? I am quite surprised and very impressed! Not sure I have seen that much DR even on a 2" master tape...

 

[Bruce B]

You have vinyl LPs with >120 dB dynamic range? I am quite surprised and very impressed! Not sure I have seen that much DR even on a 2" master tape...

Don.... I was quoting Keith O. Johnson..

 

[Mike Lavigne]

Don.... I was quoting Keith O. Johnson..

coward!