Fear of a 24-bit Planet: High-Resolution Audio

He was worried. My friend, you see, was worried.

He was worried because he’d been listening to some high-resolution audio files and found he preferred them to standard CD-quality. He called me and asked if I could come over. †

“What the hell am I supposed to do now?” he lamented as we sat in his living room.

I shook my head with a look of regret and sighed as I slumped my shoulders in sympathy to his plight.

He fretted and rocked back-and-forth on the sofa as I sipped my beer. Several minutes passed.

“You could do needle drops,” I said to break the silence.

He turned his head to look at me – a flash of anger – then we both burst into laughter. Removing his glasses and wiping the tears from his eyes a moment later, he sighed. “Maybe it’s not that big of a difference?”

I nodded and smiled reassuringly. Quietly afraid a bit for him since he had just spent the better part of a year ripping several thousand CDs to a music server. Celebrating this feat finally being accomplished he’d gone and bought several 24-bit/192kHz downloads of some of his favourite albums to add to the drive while he was into a bottle of Malbec, and well, now this.

I’d seen this sort of thing a fair amount in the last couple years. Forums online, conversations at hi-fi shows in Chicago, Munich, Los Angeles… people emailing me. Many out of sort over how they thought 24-bit files sounded superior to 16-bit. You’d think they’d be happy, I mean, they found something they liked – not really a normal state of affairs for audiophiles. But, instead they were all worried because they had so many CDs, or had, like my friend, spent the better part of hundreds of hours over several months ripping massive silver-disc collections to hard drives to play wirelessly to their new streamer/DACs. But along with that new found freedom also came the ability to stream off the Internet and its myriad services. A Pandora’s Box with untold numbers of MQA titles and 24-bit/192kHz files.

Suddenly, Redbook seemed anemic for many in comparison to those fat high-res albums, despite the fact that many used TIDAL where 16/44 is de facto unless MQA is involved. Qobuz also has a lot of Redbook, but their high-res offerings are growing exponentially, and with Amazon Music HD and Ultra HD (24/192) bullying its way into the audiophile music-streaming market, I imagine it will only get worse for some as it gets better for the rest of us.

The thing I don’t understand about high-res envy is that I don’t necessarily find it’s always better than Redbook, different many times, yes, but not always better. Personally, I thnk there’s more than enough data in a 16/44 file to provide sonic spiritual enlightenment and if the DAC you have isn’t cutting it at that resolution, what makes you think 24/192 would be better?

Perhaps it depends on your mindset and what your CD investment is looking like. I’ve only ever had a few hundred CDs at a time, I guess it’s my age – I was a bit too young to have enough disposable income to amass an atomic-weight sized hoard. I wasn’t one of those collectors with thousands or tens of thousands of albums. Perhaps with that level of buy-in I’d be singing a different tune. Perhaps not. I’ve always tried to limit how much material goods hold sway over me, and having either a CD or LP collection that requires more than a dozen boxes to move just wasn’t ever in the cards. If I don’t play something for a year, out it goes.

Those that follow my ramblings here know I have heartfelt joy for high-res audio – as well as DSD and MQA – all flavors really. It’s just like vinyl though, it really depends on the pressing, so to speak. Provenance plays a huge role in the quality of mastering an album either in analog or digital. Wehre did that file come from? What was its source? I’m down with any mastering that isn’t taking 16/44 and doing the up-rez. Same goes for LPs, why bother cutting a lacquer from CD quality? It’s vagaries such as these that keep me wary when people say 24-bit/192kHz must sound better than 16-bit/44.1kHz. As with everything in high-fidelity, the proof is in the listening. And never forget, that listening is entirely subjective; what I love you may dislike and vice-versa.

The point of all this? Try to not get too caught up in the bleeding edge of the hobby. †2 Don’t be afraid to let the spray from the bowsprit get your shirt wet, but try to keep one foot back on the deck where it’s dry lest you think the situation is one or the other. It’s usually all things at all times and taking that in stride to plot a course that keeps fun at the forefront and worry in the rearview mirror should be of paramount importance.

Life’s too short to fret whether something might be better – enjoy what you have and learn to enjoy exploration, not fear the consequences of it.

Remember, getting there is half the fun.

†The character referenced in this piece is fictional and was created to illustrate a point. †2Mr. Arnott spends far too much time and money involved with the playback of recorded music and hopes his mistakes can be learned from to spare others his suffering.

COMMENTS
Dick James's picture

Try Mark Waldrep's The High-Resolution Audio Challenge, http://www.realhd-audio.com/?p=6197. When the mastering is not a factor, you may find there is no difference. Mark Waldrep runs AIX Records, a hi-rez music recording house and store.

jim tavegia's picture

I record everything in my studio at 2496. I have tried 24192, but I can't always hear the diff, but at 72 that is my problem and not the file size. Jitter become more of an issue the higher the sample rate is.

In my sound editor if I take the 2496 files and make new copies to redbook, you don't see much if any change in the waveform going down, but when you go from 96KHZ to 44.1KHZ you can see a dramatic change in the shape of the waveform. It is very obvious. There is also a sense of smoothness gone in comparison.

I will also say that the files seem to always sound better starting at 2496 and reducing them rather than just recording them at redbook. I am guessing that my Sony SoundForge software does a great job at reducing the files. When I do this I keep the 2496 files and just have the redbook as well for a comparison anytime I want.

I would love to go DSD with a Tascam DA-3000, but I do not know of any affordable software for my computers. A Sanoma is out of my price range. lol But, that being said, starting out with DSD and then just making a analog transfer to redbook may sound better than my going 2496 to start.

It is a great time to be an audiophile and have home studios. High bit rate USB interfaces for I/O are really affordable now.

Dick James's picture

When you downsample, what type of filter are you using? Does it have a linear or minimum phase shift? I've used foobar2000 and Audacity for this purpose, I have not used SoundForge, and there are many filter adjustments that can be made. Those settings are part of the "mastering" by converting the music source format to the delivered format, whether it be files, CD, LP, etc. Upsampling has the same adjustments.

jim tavegia's picture

that is not mastering per se. The goal is to find what you find YOU like the best based on your gear, amp and playback speakers or can. It is all going to be different for everyone who records. I'm sure I don't use the same DAC as you and each of my 4 Tasdam digital recorders use a different internal AD/DA I'm sure. I do regret that I don't own the same gear as Greg Calbi, but life goes on.

ednaz's picture

There's absolutely a difference between 24 bit and 16 bit. Observing and listening to guests while we've got music playing, the increase in bit depth is WAY more impactful to people than the increase in sample rate. 44K to 48K differences are the sound of angels dancing on pin heads, but 44 or 48 to 96 almost everyone can year. Interesting to me to listen to guests talk... bit depth seems to have the biggest impact, but the sample rate jump to 96k is noticeable to everyone except the former artillery commanders or deep sea divers.

The 96 to 192 jump is definitely more subtle. Despite my first rock concert being the MC5 (who I saw several times) and being a big fan of some of the loudest bands of my era, my hearing is still unusually good. It's only in the last year I stopped hearing the CRT scanning in checkout systems in stores, but I still hear well enough to enjoy the difference in sample rate. Seems to me that enjoying the 96 to 192 difference is not just subtle, but limited to those with well preserved top end hearing.

I do still buy a number of albums in 256K or so lossy compressed. All of them bands that have totally embraced DR compression, with a max DR of 6 or less. I can't tell bit depth or sample rate differences if the music's a casualty of the compression wars. I've also seen interviews of electronica artists talking about their samples, and a fair number work with Apple 256K level samples... so why buy anything more? (since I still enjoy their compositions...)

We do live in exceptional times.

Dick James's picture

If you take a 24-bit file and convert it to 16-bits yourself, I think you'll have a hard time telling the difference. If you are using a 24-bit track from HDtracks, for example, and a CD for the 16-bit track, you are hearing the differences in mastering, not bit depth. You could also try this for sampling rates as well.

EdAInWestOC's picture

So many younger people full of self confidence in all of the knowledge they have. Then suddenly they get exposed to high resolution audio or, god forbid, analog audio.

The moment happens when they discover they weren't really listening to everything there was. Their ears have an epiphany and whack, they suddenly discover they really don't know everything.

First off only an idiot thinks they know everything. Intelligent people know they are just scratching the surface no matter how deep they think they get. We are better when we don't think we know everything. We have a better opportunity to learn and there is always something to learn.

The difference between high resolution and Redbook is the extra data to allow the DAC to more accurately reproduce the original waveform. How audible this can be depends on the performance and how complex the waveform gets.

When its right the high resolution version sounds more relaxed and lifelike. Our ears don't like unnatural sounds and digital audio can create some moments where we want to get up and leave the room.

It is why so many digital audiophiles search for the best version of a digital performance. And, like it or not, the best and most lifelike version of a performance is on a full analog reproduction.

Yeah, I know analog has its issues but when it is right there is nothing close to its beauty. High resolution is a very close second and Redbook is down the chain with barely enough data to reproduce simple high frequency content.

Redbook can be good but I find it interesting when audiophiles promote a Redbook performance and it is usually a simply jazz quartet or a single instrument acoustic performance. Then Redbook can hopefully get the performance right.

I own a lot of 24/192 files. I listen to them via headphones while I work. I enjoy them and they do a pretty good job. My main rig is all analog.

I thought I should say that as a sort-of disclaimer.

Enjoy your tunes,
Ed

Dick James's picture

...in hi-rez vs. CD resolution is really differences in mastering or the A/D converter that was used in the many transfers from tape if the master recording is analog, and that is why that person should try the Audio Challenge because it removes those variables. CD resolution is 16-bits, a 96 dB dynamic range, which is more than any music produces; classical music has a dynamic range of 40 dB at best. Using 24-bits during recording and mixing makes sense, but delivering that to customers doesn't make sense when 16-bits is plenty. The higher sampling rates make more sense only because the filters used with the A/D converter and then the customer's DAC can have less phase distortion by not needing to be a brick-wall type, but even when good filters are used, 44.1 kHz is plenty good and doesn't need to be higher. Take the Audio Challenge and you will understand that what most people are hearing are differences in the remastering process because apples are being compared to oranges.

jim tavegia's picture

I don't remaster between or after taking my 2496 files to redbook in Sony SoundForge. There is definite loss going to 44.1 from 96. if every bit is 6db of dynamic range there really is not much of a point fretting about 96db of dynamic range vs. 144 (24 X 6).

My noise floor in my home studio is now down to -80db, and I have many CDs when put in my computer that are more like analog tape at -60db or -65db in the silent passages. Even Telarc didn't have 96db of dynamic range on their music, and they were a great recording company.

The noise floor will always be the problem for anyone recording. Using a "gate" may take the noise away, but that is removal, not reality. Once you do live recording in a venue and you open them mics, you can easily hear the inherent noise of the venue, mostly the HVAC system and occasionally traffic outside the building. I have mics that have only 5db of self noise so they are not the problem. Most mics are in the range of 17db to 25db of self noise, even the great ones.

ednaz's picture

I've made playlists for dinner and evening listening parties, and have had a few cuts be halfway through at Redbook and other half at 24/96 or better. No pattern to which was first. Yep, there was a brief gap that signaled something changed. But other than my friends who commanded howitzer brigades or battle cruiser artillery, every person - even in the dining room, one room away from the system and speakers - has asked why the first part (or second part) sounded so much better than the other.

90% of my wife's listening is books on tape stuff. But she can tell if we're listening to our 24/96 version of something, or the lower quality level from streaming.

Saying it's impossible for other humans to appreciate the differences between 16/44 and higher bit depths and sample rates is claiming that whatever you can sense is the species standard. I can accurately tell you, hearing two versions of a song, whether the sample rate or bit depth increased, and I'm 64 years old. (Only in the last year have I stopped being able to hear CRT scanning in terminals and cheap old TVs.) With very well recorded music, I can tell you 48K versus 96 versus 192 with great accuracy... on my one very high end system. Not on a couple others I have that most people would consider to be very good.

I have friends who can take two mouthfuls of a wine and tell you what grapes, what part of the world, which vineyards, and what year, with accuracy a tick or two under 100%. (I've watched their certification tests.) I can reliably tell red from white, and merlot from cabernet... and that's it. Human capabilities are not evenly distributed.

My former artillery commander friends spend a lot less on gear and music than I do, buying stuff that matches their hearing. How great for them that they've got the hearing to enjoy music! That I can hear something different from them? Two of them can do the sip wine and tell the entire lineage. We envy each other from time to time.

ednaz's picture

Everything you're saying is based on some kind of statement of "average" human perception. But as a quant, I know very well that averages represent something that doesn't exist. There's no average human.

Saying no one can hear the difference between RedBook and higher definition music is like saying, we can all sip a glass of red and tell which grape, region, vineyard, and year. There are people who can do that. (I know a few who every few years go back for recertification and are able to correctly ID that string on 9 out of 10 glasses of red or white wine.) Like saying any of us could drive like like Lewis Hamilton, if we just had access to a Formula 1 car. When I raced cars (and I sucked, btw) I could tell if a tire was 1 psi off target. Good racers, like F1 or IndyCar can tell fractions of a psi difference. They can tell if the front wing on a car is a half degree angle off where it should be for maximum downforce versus speed.

I used to be able to tell you what metals were in the bell of a trumpet or trombone, just by listening to them play. I'm dead sure most people can't do that. I'm also dead sure that a lot of people CAN tell the difference. You won't find yellow brass and silver bells mixed in any orchestra on this planet - because people in the profession can tell the difference, and it matters to them.

Don't get me started on "blind listening tests". That's an easy shibboleth to demolish. But... "convert back to 16/44?' I've done what I consider to be a very indicative test. I've had playlists on in our family room system, which is one room away from the dining room (but open to it) where I'll have the same song in the list in two different places, once at 24/96, 24/192, or the 44 variants. Guests get up and to into the other room, even if in the middle of a steak cooked to their preference, when the higher def music shows up.

If we were all average, Simoe Biles wouldn't have a half dozen floor routines named after her. Lewis Hamilton wouldn't be a multiple F1 champion. There wouldn't be sommeliers because we'd all be able to taste and distinguish what they could. I spent 15 years training my ears. I suspect that means I may be able to hear things better than most.

To declare that no one can hear what you can't hear, or what a mythical average meat puppet could hear, is to deny that one can, through hard work, and better than average sensory capabilities, outperform someone picked at random off the street. I always tell people, careful what you wish for... or your doctor may be no better at detecting melanoma than your office mate.

Don't deny that sensory capabilities aren't evenly distributed, or that training can increase sensory skills. Otherwise you're denying expertise.

Dick James's picture

hi-rez releases vs. CD. These will almost always sound different. When you compare the hi-rez release that you or someone you trust has converted to 16/44.1 with a properly configured filter, most people will not notice a difference. What you are hearing is the differences in how music formats are mastered for release. Hi-rez files are the latest marketing gimmick, but even I have to buy them because CDs are mastered so poorly with so much extra dynamic compression. Compare apples to apples (I don't care about grapes) rather than apples to oranges, which is what you are doing.

jim tavegia's picture

The sample rate not the bit rate. Think of a loaf of bread looking at the side seeing all the visable slices and going from 10 slices to 20 slices passing by in one second and being able to follow the height and the low with each slice. Then go to 44.1K to 96,000 slices per second and we can now more closely follow the analog wave form that becomes a rather bezzare shape with music (you can see it by stretching out the wave form in your software) and being able to pick up every rise or fall of amplitude, now think of 192,000 snapshots per second or with DSD at 2.8 million or 5.6 million times per second...the most accurate of all.

I have recorded at 16/44.1; 16/96; and 16/192 and done the same with 24 bit and it is always the sample rate that wins in the end. Once you get to 16 bit and then count on very low jitter in the data stream as most DACs are now,

I doubt that is is that much more costly to press DVDs than it is CDs, so the industry could have been supplying 2496 music as DVD-Vs, playable on most if not all DVD players except those from Pioneer and Yamaha who chose the DVD-A format. MY discs that I have burned as DVD-Vs sound excellent even on lesser DVD players. My non-audiophile friends could even hear it. I do not "master" them to as to try and trick them. That is pointless. It is a recording left in its raw state.

Dick James's picture

A DVD-A player looks for the AUDIO_TS folder on the disc and plays that first if the DVD player is configured to do that, otherwise, it plays the VIDEO_TS folder. Which folder is seeks first is in the settings for the player. I have 4 DVD-A players and they all work this way.

By the way, DSD is a 1-bit format, so the sampling rate has to be high; the PCM equivalent sampling rate for DSD64 is 88.2 kHz.

Doak's picture

...then stick with it.
I find the "24 bit conundrum" only amusing.
After listening to excellently recorded or remastered 24/192 files there is no going back - then there's DSD64 and DSD128.
Of course really good analog playback has its own strengths that's still, IMO, the gold standard.

Dick James's picture

...in 24/96, then converting them to 16/44.1 will sound nearly the same. Take the Audio Challenge and see how you do. The Audio Challenge was designed by a person who runs a hi-rez studio and sells hi-rez recordings, but there was no bias in the test's design, and I think even he was surprised by its results.

Dick James's picture

I can believe there is a difference between a recording made with a 24-bit depth and a recording made with a 16-bit depth of the same performance. If a critical listening test is done between the original 24-bit recording and a copy of that recording that was converted to 16 bits, few critical listeners will notice a difference, and even fewer will identify which is the 24-bit file and which is the 16-bit file correctly, assuming, of course, the original recording does not have more than 96 dB of dynamic range. For consumer delivery, 16 bits makes sense; anything higher is a waste of storage space.

I have many 24-bit files that sound better than my CD of the same album, but that is only because the process of getting the music from the source to the CD, usually generically called "remastering" by the industry, was not the same as the process of getting the music from the source to the hi-rez file. There were differences in EQ made during mastering, different ADCs were used, etc. I'm finding that new releases or reissues have even less difference between the hi-rez file and the CD because the hi-rez file is being used to master the CD and the EQ changes and dynamic compression are being made to the hi-rez file. In the past, the EQ and dynamic range compression were added at the CD mastering stage so that the hi-rez file sounded closer to what the master source (I'm assuming, I've never heard a master) sounds. Check the Dynamic Range Database and you will see this trend.

Doak's picture

I trust mine.
I've listened to THOUSANDS of h-res audio albums going back 10+ years.
Well done truly hi-res stuff SOUNDS BETTER.
I use an analog tape analogy to give folks something possibly more relate-able. 16/44.1 = 7 1/2, 24/96 = 15IPS, 24/192 = 30....
Yes, I know someone out there wants to show how this is not a "fair" or valid comparison .. save your breath and time.
MORE INFO = BETTER SOUND
It really is simple. Why throw away information???
What's the cost? Disk space is cheap.
Rant away ,, I'll waste no more time on this one. 10 years in folks and TRUTH is clear.

Doak's picture

"I have many 24-bit files that sound better than my CD of the same album, but that is only because the process of getting the music from the source to the CD"

Dick James's picture

...for various formats is a well-reported topic, including by Stereophile.

X