Is greater than 24/96 worth buying in to from a future proof point of view?

[bmckenney]

I realize higher resolution digital is available today from a select few places and from a sampling and bits perspective and I know it sounds better. If I was to buy a DAC today should I aiming for greater than 24/96 because it is very likely that mainstream digital music will be available in higher resolution? Or should I be safe for many years with 24/96? If this has been discussed here before perhaps someone can provide a link to the thread. Thanks.

 

Bryan

 

[cfmsp]

First off, there's no right or wrong on this topic...you spend your money and takes your chances.

 

"If I was to buy a DAC today should I aiming for greater than 24/96 because it is very likely that mainstream digital music will be available in higher resolution? Or should I be safe for many years with 24/96?"

 

Let's see here, how much actual music is currently available at even 96k, let alone above it? Practically nothing.

 

FWIW, I'm not counting the precious few recordings done by the audiophile labels as 'actual' music, for the purpose of this discussion. The mainstream labels have provided next to nothing to date. Indeed, mainstream audio seems to be moving AWAY from hi-res to a downloadable MP3 model. I'm not talking just iTunes, but Amazon, etc. as well.

 

As for whether mainstream labels will re-master music, in order to sell it at a premium, at 96k - that's a different discussion. OTOH, there is NOTHING they can do that you can't (via use of iZotopes SRC in Wave Editor, for example) if it hasn't already been recorded at bit rates / depths higher than RBCD.

 

If it has been recorded at 88.1 or 96k, then great, they can release it at a higher quality level. BUT...you can still listen to this on a 96kHz-capable DAC.

 

You see where I'm going with this, right? We're still wondering how soon we will receive benefits from having purchased a 96kHz-capable DAC.

 

IOW, We've not gotten the industry to make the move to 96KHz, other than the aforementioned specialist recording outfits.

 

Why worry about future proofing oneself by purchasing a 192kHz capable DAC now?

As Barry Diament has pointed out - it takes an exceptionally well done 192kHz DAC to even benefit fully from the increased resolution.

 

I've put my money where my mouth is - I have two DACS that are limited to 96kHz - the Metric Halo ULN-2 and the Wavelength Proton - the latter of which I just bought.

 

No doubt others will have even more cogent reasons for charging ahead now with 192k DACs, not the least of which will be the 'improved' sound quality possible.

 

To each her/his own, but I stopped buying 'audiophile music' - just because it sounds better - years ago.

 

So, yes, music may eventually move up(stairs) to higher resolutions - but I don't think it will be leaping beyond 88.1 / 96kHz any time soon. .. at least not unless mainstream companies have been recording (behind the scenes) at 176 / 192 sample rates for quite some time already.

 

 

YMMV,

clay

 

 

 

 

[Andrew S.]

The reality is 99% of my music (and everyone else's) is 16/44. I have some 24/98 and 88 and 20/44 and 20/48. But not much.

 

Most of the great recordings are on CD - very few "audiophile" releases I actually would want to listen to....some Linn Classical aside.

 

Not saying I agree with it but I can't see the mass market wanting 24/96..

 

I have to say I often have trouble telling apart 320kb/s from Lossless.

 

[davidR]

I don't think so personally. It's currently hard enough to get 16/44 content online for downloading. I've downloaded a few songs on Itunes and after comparing them to their redbook versions stopped paying premium for a file of less quality than the album itself. To me it's a bit insane paying $10 for a "lesser" quality compressed album when the "real" version is $12. BUT I really long for downloadable redbook content because I completely understand the convenience factor. Just my take on things but I've spoken with several recording engineers around Nashville and they just laugh at the thought of consumers wanting hi-res. It's quite a struggle already to help the musicians themselves who make the music dig the concept of hi-res, not to mention the majority of people BUYING it can't hear the difference in an mp3 vs. an AIFF. We have such a "loud" standard as audiophiles yet such a "quiet" voice. Who's gonna listen to us when we are a mere percentage of the market? Don't get me wrong because I support higher quality formats period. But first we need to focus our energy on getting a majority of redbook available online & downloadable in an uncompressed format and forget all this "demand the industry and record labels to go hi-res"; because with that mindset we haven't simply jumped the gun; we have already crossed the finish line without even listening and then wonder why everyone is staring.

 

[ciamara]

A lot of studios in the US still record at 24/44 and 24/48 to minimize drive space usage, processing overhead etc. In Pro Tools, higher track counts at high res require several HD Accel cards and fast hard drives.

 

That said ... Sony is converting their entire catalogue of analog master tapes to digital. They will be using high end ADCs and recording at 24/96. So hopefully this is a sign of things to come.

 

We audiophiles are sadly a minority .... Mainstream consumers choose convenience over quality every time. We saw that with CDs, then MP3s and the iTunes Store. However, I do believe that as the speed of the internet increases and the cost of storage decreases, it will be easy and cheap to offer downloadable music to the masses at a resolution we audiophiles want.

 

Perhaps we should start a lobby group to accelerate this change!

 

[watchnerd]

Be honest with yourself:

 

If blindfolded, can you actually and truly tell the difference between 24/96 and 24/>96?

 

If you can't, why bother?

 

Those 24/192 files are freaking huge and, IMHO, massive overkill for most people and certainly offer diminishing returns relative to 24/96, even among those who can be proven to hear the difference.

 

 

 

[jonmarsh]

In the interests of equal coverage, I'll chime in that unless you have plans for specific material mastered at 176.4 kHz or 192 kHz, 24/96 really should be pretty OK- yet I've held off on getting a Logitech Transporter because it maxes out at 24/96; otherwise, it's an interesting piece of kit at the price.

 

Now, with a 176 kHz sampling rate, we have a maximum of 20 data points at 20 kHz to describe the sine wave- doesn't seem like overly much. However, if you subscribe to the concept that maybe system clean bandwidth to 40 kHz might be desirable at times (if you've never heard the difference in IM and clarity between an Accuton diamond tweeter versus their ceramic version using the same motor, you might not subscribe to that notion), then at 176 kHz we're only looking at 4 samples per sine wave, which doesn't seem overly much for PCM, especially not compared to what DSD does...

 

As Chris says, there are many paths to Rome- pick the one you like and feel comfortable traveling! ;^)

 

[ciamara]

Specifically w.r.t. the final audio mix (i.e. the song that end users buy and listen to), I notice a big step up from 16bit/44kHz to 24bit/44kHz, and a smaller, but detectable step up from 24/44 to 24/88 or 24/96. So far, the move from 24/96 to 24/76 or 24/192 seems marginal. Sure, most of that very high res material sounds great, but that is because the mix and mastering engineers spent a huge amount of time and energy recording and mixing properly with attention to sonic fidelity.

 

I often fool people (to prove a point) -- I will play a series of high res tracks and then throw in a redbook track that has been mixed and engineered well. Listeners are always shocked to learn how good some redbook material can sound.

 

So a lot comes down to what the engineers do. Do they care about the audiophile listener, or are they targeting mainstream consumers that listen to radio (or to bring us into the current century, iPods). A fun thing to do is to track down copies of your favorite albums from other markets, such as Germany and Japan. For many of my favorite artists, I have been able to find Japanese masters, and the improvement in sonic fidelity and dynamic range is truly impressive.

 

The point I'm trying to make is that mix engineering is a huge component, and that higher resolution -- if used properly -- can yield spectacular results. For example, I listen to Claire Martin, produced by Linn in the UK, and I must say the sonics on the 24/96 albums are truly spectacular.

 

To address the theoretical question of whether sampling rates of >96kHz result in improved sonics, one has to design a proper experiment. I'm simplifying a little for the sake of brevity (if I can credibly say that in a post this long!). The following should hold:

 

1. The playback system should be sufficiently resolving and detailed. The source and DAC should be well configured and able to handle and play material up to 192kHz properly. The amplifiers should not have filtering in the high frequencies (many do in order to eliminate RF noise). The power supply should be clean. Cabling should be top quality and matched to the system. The loudspeakers should have a frequency response well above 20kHz. We do our listening tests on a 4-way active system with TAD drivers, including supertweeters that have a frequency response from 5kHz to 45kHz).

 

2. The material must be recorded at high resolution by one of two methods (at least, these are the two that come to my mind -- there are surely other ideas out there):

 

2a. Record and mix in the analog domain. Then transfer the stereo mix to digital at a few different sampling rates from redbook to 24/192.

 

2b. Record in the digital domain at the highest sampling rate desired (e.g. 24/192). Compare the final 24/192 mix to 24/96 and 16/48 mixes (notice that 96 and 48 are even divisions of 192 -- this is important). In generating the lower res versions, one must use dither -- the same technique engineers will use when mixing down. My logic here is that many engineers work this way, so we should do as they do for this experiment.

 

3. Compare them in a blind listening test, by two methods:

3a. Fast switching between them

3b. Extended listening with each

 

I see on the home page of CA that there are some tracks from FIM at different sampling rates. What is not clear is the method by which these were generated. It would be helpful to know that. Armed with such information, and assuming the different versions were made with one of the above methods, we can all go back and listen to them in blind comparison tests and see if we can pick out the difference.

 

Certainly, I would like to do that ... Anyone else?

 

[barrows]

Here are my thoughts: Most stuff that is recorded with care today is done at 24/96 and archived as 24/96. Some stuff is done at 24/192, and some at 24/176.4. During the SACD mini boom, a lot of classic titles, rock and jazz, was re-mastered to DSD, especially Sony titles. This stuff is most likely all archived as DSD. Additionally, most classic titles that have been re-mastered since the mid nineties (from analog master tapes) have been done at 24/96 or higher. So, leaving out classical music for a moment, the record companies generally have archived files of most classic titles in Jazz and Rock music in DSD, or 24/96. Most mastering engineers believe that single bit DSD converts to high res. PCM very easily. Considering that many titles are archived at high resolutions, it is not out of the question to believe that some, or even all, of this music will be available in the future, at resolutions of at least 24/96 in some format. For the above discussion I am talking about popular rock and jazz titles. Personally, now, I have 24/96 music from: King Crimson, Nine Inch Nails, Robert Plant & Allison Krauss, Fleetwood Mac, Neil Young, Talking Heads, etc.

If you never expect to appreciate, or ever listen to, Classical music, I would suggest that playback of resolutions above 24/96 will be nearly useless, as I really doubt there will ever be more than a handful of titles of Rock and Jazz available higher than 24/96. But, if you do appreciate Classical music at all, there are already a lot of classical titles available at 24/176.4 and 24/192, all you have to do is search around a little-I also would not classify this music as "audiophile", the music is great classical music.

Can you hear the difference? That will depend on the original quality of the recording, and the resolution of your own system. Right now the very best DACs can only resolve about 20 bits of dynamic range (of course, this is substantially better performance than 16 bit CD) and very few home audio systems have any more than 120 dB of actual dynamic range. Sampling rate is a different story though, it is likely that very high sampling rates (up to 352.8) will result in audible improvements (if you have a DAC that is well enough engineered to do this high rate justice) if transient performance equal to (or better than) the best analog tape is expected.

For me, I want 24/192 capability, as there are classical titles available of music that I love at resolutions above 24/96. If one never listens to classical music, then you probably will not miss much if your set up is only capable of 24/96, but I would not recommend investing in any gear now that cannot handle at least 24/96.

 

[cfmsp]

 

"Sure, most of that very high res material sounds great, but that is because the mix and mastering engineers spent a huge amount of time and energy recording and mixing properly with attention to sonic fidelity."

 

Totally agree with Sanjay here, and I think this is a point that many miss when they consider the need for 192k capable gear.

 

As an example, I have an MP3 which sounds particularly good to me, even at AAC sampling rates. I do NOT notice any obvious drop off when listening to it.

 

Of course, it might sound even better at higher sampling rates, BUT, the point being...a well done recording elevates the perceived sound quality (perhaps) more than a higher sampling rate.

 

clay

 

 

 

[barrows]

that this really an apples to apples comparison: of course a well done recording is going to sound better than a badly done recording. If one wants to find out if they prefer higher resolution in their system, it is pretty easy to test this. Find some music you like, that is well recorded, and available on CD and at 24/96 or higher. Playback the two versions in your system and see if you can hear the difference. For me, Robert Plant & Allison Krauss' "Raising Sand" is perfect for such a comparison, a decent (but not perfect) recording, and it's available at 24/96 from HD tracks.

From a technical standpoint, consider that the requirements of the digital filter are relaxed considerably with a higher sampling rate; this means that the sonic degradation caused by the digital filter can be moved well out of the audible range with higher sampling rates. Theoretically, even a mediocre recording will benefit from a higher sampling rate, because a less intrusive digital filter can be used, resulting in a reduction in digital artifacts in playback.

 

[cfmsp]

"From a technical standpoint, consider that the requirements of the digital filter are relaxed considerably with a higher sampling rate; this means that the sonic degradation caused by the digital filter can be moved well out of the audible range with higher sampling rates. Theoretically, even a mediocre recording will benefit from a higher sampling rate, because a less intrusive digital filter can be used, resulting in a reduction in digital artifacts in playback."

 

Well, to separate out the improvement in sound due to higher quality recording - which was Sanjay's point - wouldn't you just upsample (with say Izotope) to gain the benefits you describe?

 

How big ARE these benefits, as compared to say the improved recording quality?

 

Just to be clear, I'm not suggesting that there are not benefits to 96kHz playback, just trying to sort out the possible combinations and discover which makes the most difference.

 

1) recording in 88.2/96 versus recording in 44.1/48

2) playback in 88.2/96 versus playback in 44.1/48 (of a 96k recording)

3) playback in 88.2/96 versus playback in 44.1 (of a RBCD recording)

 

Probably #1 has the most impact - that was the implication of my earlier comment - a good recording matters most.

 

Probably there is a descending level of improvement for the three options outlined above.

 

It seems like what you're describing above (in the words I quoted) is #3? Perhaps I misunderstood - wouldn't be the first time.

 

clay

 

[The Computer Audiophile]

This is a good discussion I just wish we had a few very good engineering opinions. I know two engineers may not agree but at least it would some additional info for us. The reason I say this is because one of the most respected recording engineers in the world has described why 176.4 is beneficial / required and show graphs to illustrate his point. On the other hand there is Dan Lavry who I believe states a sample rate around 70 kHz is all we need.

 

It's not something simple for us armchair engineers.

 

[barrows]

We have no control over the quality of the original recording, we are stuck with it. If there is music I like, I will listen to it regardless of the recording quality, but, if I can get the same music in a higher resolution format, I am definitely going to take advantage of that.

Certainly, I am not advocating that crappy recordings that are mastered at 44.1 or 48 K should be upsampled to 88.2 and then re-sold. I suspect that there could be some benefits to upsampling oneself using Izotope, but there is a tradeoff as well, you are now subjecting the file to an additional digital filter that will add its own artifacts.

What I am saying is that most recordings produced since the mid nineties were made at resolutions higher than 16/44.1, and most music that originates on analog tape that has been re-mastered since the mid nineties, also is at resolutions higher than 16/44.1 (hopefully from carefully restored analog masters). Most music is now archived at resolutions higher than 16/44.1, some at 24/48, some at 24/96, and some as DSD. I think that it is reasonable to expect, that at some point, more of this music will be distributed in higher resolution, perhaps from places like HD tracks, perhaps by the record companies themselves, perhaps by the artists themselves (as NIN does). Right now, between ripping DVD-As and DVD-Vs, downloading, and (for the adventurous) ripping Blu-ray there is quite a bit of music available at high resolutions.

It would not be a valid comparison to compare a CD file to an upsampled version of the CD file, as the upsampling process subjects the music to an additional step of digital filtering, additionally, upsampling will not improve the resolution of music that was 16/44.1. As I noted in my previous post, if one wants to make a comparison, get something you like on CD, and then get a higher resolution version of it and compare both files playing back through your system. Something like "Raising Sand" is perfect for this type of comparison, as it is readily available on CD and in 24/96, and the 24/96 version was not created by upsampling.

 

[bdiament]

Hi watchnerd,

 

"If blindfolded, can you actually and truly tell the difference between 24/96 and 24/>96?"

 

With the ULN-8, easily.

I've said before, I find recording at 192k on the ULN-8 to offer greater improvements over 96k, than 96k offers over 44.1k.

 

Files are larger, yes. It depends on what one wants. The best sound I've ever heard (even beating analog in the areas where the best analog still beats 96k) has been with the ULN-8 at 192k. Enough difference, easily enough heard, that I do all my recording at 192k now.

 

It really depends on the unit. I've heard some units that don't sound as good at 192k as they do at 96k. The demands on the designer (clocking, etc.) just go up faster than many designers' apparent ability to keep up with them. With the '8, I'm in sonic Heaven.

 

Just my perspective.

 

Best regards,

Barry

www.soundkeeperrecordings.com

www.barrydiamentaudio.com

 

[The Computer Audiophile]

HI Barry - Your point about some units not sounding as good at 192 echoes what I've heard as well. DACs have a sweet spot and work best at certain rates.

 

One a side note, why do you work at 192 instead of 176.4? So much material ends up at 16/44.1, a multiple of 176.4. It's good to hear an opinion from someone who does this for a living, and does it very well.

 

Thanks Barry.

 

[cfmsp]

 

 

Thanks for the detailed clarification, Barrows, I think I get your points now.

 

I should listen to more 96kHz music, so I'm off to search.

 

Until then I'll 'listen in' on this thread, aka I'll STFU.

 

 

thanks again,

clay

 

[bdiament]

Hi Chris,

 

"Your point about some units not sounding as good at 192 echoes what I've heard as well. DACs have a sweet spot and work best at certain rates.

 

One a side note, why do you work at 192 instead of 176.4? So much material ends up at 16/44.1, a multiple of 176.4. It's good to hear an opinion from someone who does this for a living, and does it very well.

 

Thank you for your kindness.

 

As to some DACs working best at certain rates, in my experience, this is not arbitrary (for example, great at a 4x rate and not go great at 2x). It always seems to be a ceiling. Many DACs can achieve 2x rates (i.e. 88.2 and 96k) but cannot do as well at the 4x rates (176.4 and 192k).

(This seems to echo my experience that much software, though spec'd for 24-bits, can handle 16-bits but at 24, does not produce clean results in the low order bits.)

 

As to 192k, I might ask why 176.4?

The fact that much material ends up at 44.1 (of which 176.4 is an integer multiple) appears to lead many folks to believe the sample rate conversion is better with 88.2 or 176.4. While this used to be true with the earliest generations of sample rate conversion algorithms (and might still be true today for the lesser quality SRC algorithms), it is not true at all for the best SRC algorithms.

 

Alexey Lukin's superb sample rate conversion, marketed as iZotope's 64-bit SRC is to my ears, much more transparent, even at non-integer conversion (say, 192k to 44.1k) than anything else I've yet heard, even when that other is working at the "easier" integer conversion. Where others invariably generate spurious harmonics, tending to brighten and harden the sound, Alexey's algorithm just converts the rate, without the added "edge". (Though some confuse the results with "detail", they are in fact, distortion.)

 

Further, I have released material at 24/96 (using the PCM tracks of a standard, video DVD, that plays in any DVD player). In the future, we'll release files-on-disk at 24/192. Since I want the best results I can possible get, I chose not to limit this in the slightest, simply because one of the target formats -the lowest quality one- is CD. I'd rather approach things from the top down, rather than the bottom up. And with Alexey's 64-bit SRC, this is not an issue. (His MBIT+ dither/noise shaping algorithm is the other component that allows me to create 16/44 masters that preserve many of the qualities of the 24/192 version.)

 

Best regards,

Barry

www.soundkeeperrecordings.com

www.barrydiamentaudio.com

 

 

[The Computer Audiophile]

Hi Barry - Thanks for taking the time to explain why you are using 192 over 176.4. That's exactly the kind of information that's really valuable for us armchair engineers :~)

 

[watchnerd]

Barry, you said:

 

"I've said before, I find recording at 192k on the ULN-8 to offer greater improvements over 96k, than 96k offers over 44.1k."

 

I don't think there is any doubt that higher bit levels and sample rates have benefits during the recording / editing / mastering processing. Especially, relative to bits, for heavy multitrack recordings.

 

But I think we're mostly talking about the benefits of higher sample rate hi-rez (>96khz) during end user play back of final product.

 

 

[bdiament]

Hi watchnerd,

 

"But I think we're mostly talking about the benefits of higher sample rate hi-rez (>96khz) during end user play back of final product."

 

So am I.

It isn't just for recording, etc. that I hear the benefits. After all, to hear the benefits, I must play the file back. ;-}

 

With 192k on the ULN-8, I'm hearing what to me is essentially, the mic feed. Artifacts of the conversion processes (A-D and D-A) are so much diminished, I've never before heard the recording system get so "out of the way".

 

I hear something similar when listening to the Reference Recordings 4x (176.4 in their case) files. I've always loved Keith Johnson's work but have never heard it so clearly and free of digital artifacts.

 

So what we have in the 4x rates, in my view, is the potential for better sound than we've ever had before. In my experience, not all DACs reveal this (regardless of what their specs claim). But the best do and I find the results quite wonderful.

 

Just my perspective.

 

Best regards,

Barry

www.soundkeeperrecordings.com

www.barrydiamentaudio.com

 

 

[ciamara]

Barry, do you feel that recording and mixing at 24/192 rivals or betters analog tape? I was at AES with our partners (Summit Audio, Bricasti and Thrive Sound), and while walking around, I noticed a lot of interest in 2-inch tape -- much more than in years past. I know you were one of the first CD mastering engineers, and for this reason I assume you have a strong affection for digital. That said, how would you compare the two? My sense is that people who like tape use it for its "vintage" characteristics, the saturation ... the hiss ... whatever it is they are after. Or perhaps there is more to it. Anyhow, just curious to hear your thoughts.

 

[bdiament]

Hi Sanjay,

 

"Barry, do you feel that recording and mixing at 24/192 rivals or betters analog tape? I was at AES with our partners (Summit Audio, Bricasti and Thrive Sound), and while walking around, I noticed a lot of interest in 2-inch tape -- much more than in years past. I know you were one of the first CD mastering engineers, and for this reason I assume you have a strong affection for digital. That said, how would you compare the two? My sense is that people who like tape use it for its "vintage" characteristics, the saturation ... the hiss ... whatever it is they are after. Or perhaps there is more to it. Anyhow, just curious to hear your thoughts."

 

With the ULN-8 at 192k, (this is important because for my ears, it is absolutely not true of many other devices, a number of which cannot even do 24-bit cleanly, much less get the clocking for 192 right), for the first time in my experience, a number of reservations I've held for decades, where I've felt the best analog does some things better, have evaporated. (To be clear, I don't think it "rivals" the best analog, I think it clearly bests it.)

 

With the '8 at 192k, I feel I have the best recording, mixing, mastering device I have ever heard, regardless of price, format, etc. To my ears, there is no longer anything the best analog does better, in any area of sound I know how to describe.

 

As to a "strong affection for digital", being one of the first engineers to master for CD, my feelings have long been quite the opposite. (In those days, I had to stop listening to analog when I went home or I'd have been unable to endure working with digital all day.)

 

I still think 16/44 is the "cassette" of the digital world.

96k shows what is wrong with 44.1k. 24-bit (if the hardware and software can do it cleanly - most can't) shows what is wrong with 16-bit. And on the ULN-8, 24/192 is the largest leap upward in sound quality I've heard (appreciably larger for me than the jump from 44.1 to 96). For the very first time, not only does digital not create a painful experience, it has at long last fulfilled its promise.

 

I understand some folks like the "color" that can be achieved with analog (or tubes or... etc.). That isn't what the '8 at 192 is about. (Though it offers numerous options for applying "character" if one so desires.) I have described it as being the first time I feel the recorder has been effectively removed from the equation and I'm listening directly to my mic feeds.

 

Of course, as always, this is just my perspective.

 

Best regards,

Barry

www.soundkeeperrecordings.com

www.barrydiamentaudio.com

 

 

[ciamara]

Very interesting perspective. Thanks for this!

 

[bdiament]

Hi Sanjay,

 

I don't want to give the impression I believe analog is only good for "color".

On the contrary. I find excellent results can be obtained, particularly at 30 ips, i.e. higher tape speed. (In my view, the "color" only comes into play when analog is "pushed" or with the slower tape speeds.)

 

It is just that with the '8 at 192, I feel we now have a medium that can surpass the best analog in every parameter.

 

Best regards,

Barry

www.soundkeeperrecordings.com

www.barrydiamentaudio.com