World's best drummer explains why analog is superior to digital

[E-Stat]

So, everything I do gets backed up dinosaur style, via data discs, as flacs, burned to DVD blanks.

Maybe it's time to switch over to BR discs with six times their capacity. I have archived some stuff that way just because I've got a burner. And some blank BD-Rs.

The answer to all data storage concerns is maintaining multiple backups. Since my music collection fits in a tidy 300 gb, I have half a dozen offline copies via USB drives. That mainly because I've gradually added higher capacity models over the years as my largest space consumer is video requiring about 1 tb. Since the older drives run but half an hour each month with incremental changes, they might outlast me.

And I have a 400 gb microSD card the size of my thumbnail with my entire music library on it. Thinking back to the 70s when I began this hobby I could never have imagined that a tiny sliver of plastic and silicon could replace the space of 1000 albums.

 

[Bill Ferris]

With a data storage cost per album of about a nickel.

 

[DaveVoorhis]

The answer to all data storage concerns is maintaining multiple backups.

Quoted for truth.

I have multiple portable hard drives used as backups, multiple working computers used as backups, and multiple cloud providers used as off-site backups. All are kept up-to-date, and moved from old media to new media (and new cloud providers) as old media ages.

I don't rely on spinning plastic, whether CD-ROMs, DVD-ROMs, or BD-ROMs. If they're kept up-to-date and regularly tested or re-written, they're fine, but it's too easy to become complacent and assume an old disc is still usable only to find out -- when you need it -- that it isn't.

 

[Bill Ferris]

I have multiple portable hard drives used as backups, multiple working computers used as backups,

Dave, and in my case, a external 1Terabyte SSD + 2 500 Gig. Flash Drives, as well.

 

[toddalin]

I don't think that I can agree with the drummer.

I've listened to many systems reproducing Tin Pan Alley and even over my computer monitors, my Mofi SACD version played on an Oppo-95 through a Yamaha RX-Z9 receiver has a more tactile feeling than any of the others. You can go to that thread and judge for yourself.

When we used to record in the '70s, properly recorded drums would sound/feel like "someone bouncing BIG basketballs."

 

[+48V]

Thinking back to the 70s when I began this hobby I could never have imagined that a tiny sliver of plastic and silicon could replace the space of 1000 albums.

Thinking back to the early 80s, I envisioned exactly that.

Was helping a friend finish his basement when I had this epiphany. He thought I was crazy.

 

[E-Stat]

Thinking back to the early 80s, I envisioned exactly that...Was helping a friend finish his basement when I had this epiphany. He thought I was crazy.

You were crazy (at the time)!

Having Cray 2 supercomputer ($7M) computational abilities with a $40 Raspberry Pi platform with a quad core 64 bit processor was not something most folks envisioned even understanding Moore's Law.

In another thirty years (maybe 20?) we'll be saying the same about having exascale Cray Shasta capabilities for a song.

 

[+48V]

You were crazy (at the time)!

Nevertheless, I was right.

 

[Rdenney]

Kinda. Except that he stops before addressing the real world of brick wall filtering in terms of the unavoidable pre or post ringing when using a 44.1 sample rate. That’s the elephant in the room.

Higher sample rates as commonly used by higher quality recordings today takes care of that. While I have some very nice minimally miked analog recording, I would consider my best content as some of the 96/24 or 192/24 recordings that were mastered that way- as opposed to digital copies of analog or stuff that was updampled to masquerade as high rez.

Actually, he did, but in his description of how the odd harmonics superpose to create a square wave. The band limitation of 20 KHz resulted in ringing and reduced slew rate on the phase changes. The graphic animation he used to show each harmonic summing up was really quite clever, and clearly showed that adding further high-frequency sums would both reduce the ringing and increase the phase-change slope.

I don't see it as a contention between pro and anti-digital at that point, but between those who think 16/44 is adequate and those who believe more is better. He addressed that as much as he addressed the faultiness of the step-forward model.

For the step-forward model, it matters when we add a data slice between two measured samples. Simply: Does the added slice use the previous value? (That would be step-forward.) Does it extrapolate from the previous two samples? Does it average between the previous and subsequent sample? And at what precision? He didn't go into those strategies for filling in the blanks between samples. It may be that an algorithm doesn't need to do any of those things if the sample rate is high enough. But he did dispel the usual notions that 20 KHz does or doesn't play well depending on things like phase delay and aliasing, with a sample rate of 44.1 KHz. He used dithering as an anti-aliasing filter, which would make sense to anyone who does digital photography at a higher level.

Of course, the output analog circuitry will blur things a bit at 20 KHz, providing its own filtering effect at very high frequencies. Of course, what happens at 20KHz doesn't matter much to me--I can't head anything much higher than 12 KHz anyway.

Rick "who records at 24/96, and thinks that's probably overkill" Denney

 

[Rdenney]

Yes definitely. UHF/VHF (analog) would also be affected by weather.

But under normal conditions, I find digital video far superior to analog.

In the day when I was designing analog CCTV systems, we thought broadcast analog SDTV (4:3, 256-line, and 30 Hz refresh rate) was pretty durable. If broadcast standard was normalized to 60dB, we expected a signal at 30-36 dB to still look pretty good, and to be usable down to maybe 28 dB, with a 24 dB noise floor. Digital video was great down to about 30 dB--essentially as good at 58 dB. But at 28 dB, it lost data sync and fell apart completely. Analog signals degrade more smoothly than do digital signals, but they start doing so earlier. At least that's my experience.

Rick "good when it's good, bad when it's bad" Denney

 

[E-Stat]

Actually, he did, but in his description of how the odd harmonics superpose to create a square wave.

The only valid test for me - and that of recording engineers is to compare live feeds at different sample rates. Turn the sillyscope off and use all elements found in the real world of digital production including the necessary filtering to understand the implications.

Of course, what happens at 20KHz doesn't matter much to me--I can't head anything much higher than 12 KHz anyway.

Except that the pre or post ringing effects manifest themselves well below your threshold (and mine).

44/16 is a thirty eight year old standard entirely fenced in by limited technology at the time. One of my favorite recordings, The Sony release of John William's score for Star Wars - the Force Awakens, would take ten 640 MB optical discs, aka CDs to store. Or just a couple minute download today to a completely different world of computer technology.

The six core 64 bit ARMv8 processor in my iPhone has more power than the $32M water cooled Cray 2 supercomputer of that era. I have a backup of my entire digital library on a single 400 GB microSD card the size of my thumbnail.

The good thing today is we are no longer shackled to pre IBM PC technology!

 

[Rdenney]

Except that the pre or post ringing effects manifest themselves well below your threshold (and mine).

44/16 is a thirty eight year old standard entirely fenced in by limited technology at the time. One of my favorite recordings, The Sony release of John William's score for Star Wars - the Force Awakens, would take ten 640 MB optical discs, aka CDs to store. Or just a couple minute download today to a completely different world of computer technology.

The six core 64 bit ARMv8 processor in my iPhone has more power than the $32M water cooled Cray 2 supercomputer of that era. I have a backup of my entire digital library on a single 400 GB microSD card the size of my thumbnail.

The good thing today is we are no longer shackled to pre IBM PC technology!

True, but that isn’t the question. If 32/384 is better than 24/96, then will 64/1024 be better still? Where is the line? At some point, the additional detail is at too high a frequency to hear, and the noise floor is unhearably below the ambient noise of the real world. In my house, even without the air handler, ambient noise is about 40 dB. 100 dB is about as loud as is listenable in the house. I hear nothing when I power up my amp, which has no volume control, and about 96 dB s/n, even with my ear to the speaker. I hear no noise in headphones when there is no music, even with the volume at a level that would cause damage.

Maybe someone needs a noise floor 30 or 40 dB greater than the listening environment, or the source material, but not me.

On those ringing notch filters—do we need filters with that much brick wall? Are the resonances we are attenuating that sharp? Not in my experience. The square wave test is interesting because it is so challenging compared to just about any music material.

For weak-signal radio reception in a noisy band—that’s when I want a 130 dB s/n and the brick-wall filtration needed to keep off-frequency signals away from it. My old non-digital HF transceiver has 32 poles of filtration if I need it. But fidelity isn’t the point for that setup.

Rick “suspecting 24/96 is abundant, even if we can do more” Denney

 

[E-Stat]

True, but that isn’t the question. If 32/384 is better than 24/96, then will 64/1024 be better still?

I'm not aware of any such format. DXD is 352/24.

Where is the line?

At the point which the live feed from a symphony is indistinguishable from the recording.

At some point, the additional detail is at too high a frequency to hear

That is the popular canard voiced by those who don't understand the problems created by the necessary filtering. It has little to do with capturing ultrasonic content beyond that which we respond to. You'll note I didn't say hear.

On those ringing notch filters—do we need filters with that much brick wall?

With 44.1 most certainly. What is not absolutely filtered renders as 100% distortion.

It is a poor compromise borne of 1980 era technology. Even Dr. Stockham's Soundstream format used by Telarc et. al. was 50/16 and that was immensely impractical given that it required a Honeywell data tape drive and could only store small segments of an album at a time.

Why not leverage dirt cheap current technology to employ the best sound quality?

 

[wyn palmer]

I'm not aware of any such format. DXD is 352/24.


At the point which the live feed from a symphony is indistinguishable from the recording.


That is the popular canard voiced by those who don't understand the problems created by the necessary filtering. It has little to do with capturing ultrasonic content beyond that which we respond to. You'll note I didn't say hear.


With 44.1 most certainly. What is not absolutely filtered renders as 100% distortion.

It is a poor compromise borne of 1980 era technology. Even Dr. Stockham's Soundstream format used by Telarc et. al. was 50/16 and that was immensely impractical given that it required a Honeywell data tape drive and could only store small segments of an album at a time.

Why not leverage dirt cheap current technology to employ the best sound quality?

I confess that I'm getting a bit lost as far as the thrust of this discussion is concerned.
I thought for a while that we were heading towards a discussion of the benefits of band-limited resampling/interpolation but that does not seem to be the case.

 

[E-Stat]

I confess that I'm getting a bit lost as far as the thrust of this discussion is concerned.
I thought for a while that we were heading towards a discussion of the benefits of band-limited resampling/interpolation but that does not seem to be the case.

I observed that Monty believes that using anything higher resolution for musical reproduction than 44/16 provides no benefits.

Clearly, studies and engineers from virtually all the major recording labels disagree.

Interpolate all you wish.

 

[wyn palmer]

I observed that Monty believes that using anything higher resolution for musical reproduction than 44/16 provides no benefits.

Clearly, studies and engineers from virtually all the major recording labels disagree.

Interpolate all you wish.

I'm interested in the studies. I'm only familiar with a couple and they seem contradictory. Please provide citations if you can.
As for engineers- well, we largely do what we're told and/or what is technologically feasible and what will serve a perceived "market" niche without necessarily assigning an intrinsic "value" to the effort.
Monty is clearly an excellent engineer and his opinions differ from yours (and to a much lesser extent mine) and that's fine.
For a while there I thought that you were converging on an explanation for why Red book is unacceptable but I'm not sure any more. Are you saying that the problem lies with the anti-aliasing filter or with the reconstruction filter or both or what? And whichever it is could you clarify as to what the issue is.
Thank you.

 

[MannyE]

I dunno guys...

Musicians hear music differently.. he is clearly stating that he can hear the analog "signature" coming through even on crappy iphone speakers on a recording that was originally recorded in analog and clearly says that digital has come a long way. A long way in being able to reproduce the analog signature. He is NOT so clear about digital recordings sounding as good as analog recordings. HOWEVER... he is also very clear (at least to me) that the care that was needed at every stage of an analog recording resulted in a better end product. This is IMO, independent of his opinion that the current style of mixing annoys the heck out of him. Which is yet another subject. He covered a lot in those two sentences....

This is what I got from him...

-The engineers doing the mixing back then made better mixes...the new style generally sucks... nothing to do with the medium
- A 2 inch machine with 16 tracks running at 15 ips sounds better than anything else still
- Digital is getting there... what's not clear to me is if it's getting there for original recordings or reproducing a good analog recording.

That's my take.

 

[DaveVoorhis]

I observed that Monty believes that using anything higher resolution for musical reproduction than 44/16 provides no benefits.

Not to quibble, but for the sake of accuracy, it's worth noting that in this thread we're not referring to what Monty currently believes, only what he believed when the videos were made about ten years ago.

 

[E-Stat]

I'm interested in the studies. I'm only familiar with a couple and they seem contradictory. Please provide citations if you can.

Click here for one using what I consider the best approach - recordings of the same content at different resolutions. No Rube Goldberg-esque methodology nor unsubstantiated assumptions as you find with others.

IAs for engineers- well, we largely do what we're told and/or what is technologically feasible and what will serve a perceived "market" niche without necessarily assigning an intrinsic "value" to the effort.

Or, what recording engineers do - experiment for themselves and validate what they find best.

I'Monty is clearly an excellent engineer and his opinions differ from yours (and to a much lesser extent mine) and that's fine.

Yes, he develops a fine operating system. Everyone is certainly welcome to their own opinion, but I tend to value those who actually record music for a living.

Are you saying that the problem lies with the anti-aliasing filter or with the reconstruction filter or both or what? And whichever it is could you clarify as to what the issue is.Thank you.

I'd really rather not beat a decidedly dead horse, but the problem lies with insufficient real word bit depth and unavoidable smearing of the signal due to the required slope of the filters. Both issues are easily solved using current technology.

 

[DaveVoorhis]

Yes, he develops a fine operating system.

Developed. Apparently, he hasn't been at Red Hat since 2013.