24-bit/192kHz is pointless?

[Wavelength]

Guys,

 

I have been feverishly getting ready for the show and therefore been listening to the like 100 or so high res disk we ripped for the show.

 

I really don't think listening tells the whole story. Some of the mastering on these 24 bit 88.2->192k disks, sucks. Some of the Red book stuff I have sounds allot better. But there are some tracks that really show off how much more information is included.

 

As you all know I am not an upsampling fan. I can tell the iTunes upsampler is pretty good but the downsampler (i.e. 24/192 to 24/96) is well... not as good.

 

As a musician (guitar and percussion) I do allot of work with several musician's and recording studio's in the area. Most of the work is now recorded in multiples of 44.1 as they only output red book as the final. We typically use 24/88.2 as we find this the best sounding and easily downsamples out to 44.1/16.

 

What I really don't like is the sound of 44.1 to 96 or none even multiple upsampling.

 

As for what's best... isn't more better?

 

Ashley... I think it has been proven that we cannot look at Analog or Phono in terms of S/N ratio (i.e. their bit size relative). Let's face it these mediums are more on the scale of infinite bit and sample rate. I have recorded several albums on some expensive AD convertors at 24/88.2 and while they sound good in the digital domain. They don't sound as good as the album.

 

Thanks

Gordon

 

[The Computer Audiophile]

Hi Gordon

 

"What I really don't like is the sound of 44.1 to 96 or none even multiple upsampling."

 

I agree and it is something that'll continue to annoy us as people demand the higher numbers, i.e. 24/96 instead of 24/88.2.

 

The Reference Recordings HRx material is 24/176.4 and I applaud their decision to use a multiple of 44.1 instead of the highest number. Also, it should be noted that RR made a new digital transfer from the original analog master to create the 24/96 version of Exotic Dances available on HDTracks. Very solid move in my opinion.

 

 

[audiozorro]

I remember the manual switch settings on my former Apogee Firewire Mini DAC were 44.1, 48, 88.2, 96, 176.4 and 192, but that unit was atypical among DACs. Multiples of 44.1 make a lot of sense but that’s only because we were (and still are) so tied into 16/44.1 in the first place.

 

I recently threw out my 10-year old Pioneer DVD player which had 24/96 capability. My current audio reference DVD player has 24/96 capability and the settings on my budget Oppo universal player are 48K, 96K, and 192K. So, since video trumps audio in the marketplace I believe that even multiples of 48 are here to stay as DVD blossoms into Blu-ray. SACD is somewhat of an oddity for me. I don’t quite know where the future of SACD fits in or if it has a long lasting future.

 

 

[kana813]

If Dan Lavry is right about the "optimal sample rate" being something between 60 and 70 kHz, then the highly rated Benchmark DAC-1 which uses a Analog Devices AD1896 to convert the datastream to 110kHz regardless of the original sample rate of the data must be wrong.

 

Hey Ashley,

 

Who says Optical digital is the lead that everyone prefers because it's more versatile? Did you know there are Toslink TX/RX modules that can do 24/192?

 

BTW- do the 14 screws holding the lid on your CDP make an audible improvement? ;-)

 

 

[The Computer Audiophile]

Ah yes, the ever-so-opinionated kana813 finally joins the discussion :-)

 

I have to say it really sounds like you're trying to stir the pot with the comment "Benchmark DAC-1 which uses a Analog Devices AD1896 to convert the datastream to 110kHz regardless of the original sample rate of the data must be wrong." and your question/statement to Ashley. I believe Ashley has stated several times on the site that his customers vastly prefer Toslink over the other interfaces. Plus, I really doubt he means everyone as a literal term, rather it's likely a figure of speech.

 

 

 

[Ashley James]

I can't imagine the number of screws holding our CD player could possibly affect the sound quality, though it might stop it buzzing and rattling. It's some time since we made a CD player, all our customers have bought Macs! And I didn't know there were that many screws anyway!

 

I think this discussion is flawed because no one actually knows whether what he's hearing sounds as it does because of the sample rate or Bit rate. If something doesn't sound right, it's often simply because something hasn't been done properly. It's jolly hard to know what bit of a system to blame. And people don't even agree on what sounds right anyway. I don't much like the sound of vinyl, it can really irritate me, yet I'm often surprised at how acceptable some MP3s can sound, to me they can be better than some 16 Bit.

 

This has to be an academic discussion and an attempt to put higher Bit and sample rate material into perspective.

 

As has been mentioned already, the big variable is the competence of the implementation of the whole audio chain.

 

A friend has recently tried the analogue output of his Mac Mini into a variety of different power amplifiers and got very different results with several. Into the best, it sounded really quite good and into others surprisingly poor. If you measure residuals on the output, you'll see why. Strangely the problem gets worse with compressed material. Squeezebox analogue outs can be unusable on certain makes of amplifier and acceptable on others.

I picked a Mac Mini because no one expects it to be as good as it is and because it wouldn't do for one manufacturer to mention another's that suffers similarly, but there are still more than there should be.

 

Therefore I'm inclined to think we'll not be able to agree, but I'm of the opinion that such improvements as are possible with these high bitrates/sample rates, will be small and certainly not sufficient to persuade the anti 16 bit camp that digital is worth a listen. However I do think complex recordings made throughout at 24 Bit might be slightly better.

 

Ashley

 

[mpmct]

You guys tell me, I don't know ...

 

Discussion of bits/dacs, assuming what amplification,

which recordings, which speakers, and which room acoustics?

How does one understand ... the perspective?

 

I used to have $300 monitors, I now have $4800 monitors.

They are as different as night and day, A/B/X testing would

be only for sport, the difference is so painfully clear.

 

On the other hand, one DAC to another, and I've compared

brands from circa $1000 msrp to circa $2500 msrp ...

not so clearly different. 24/96 to Redbook, in most cases,

very very hard to find an audible difference with what I can discern

are 'good' recordings. Sort of like: "I think I can!!" ;-)

 

I guess what I'm asking is: say I have a BMW 325Ci,

that, if the speed limiter was defeated, would go much faster

than what it can go with the speed limiter in place.

Do I care? Should I dare?

 

I understand that there are so

many variables in this business, but it is difficult to put

into meaningful perspective, when one considers the whole

ball 'o wax -- the entire system, including room interaction.

 

What's an amateur to do? :-)

 

 

 

 

 

 

[tfarney]

"On the other hand, one DAC to another, and I've compared

brands from circa $1000 msrp to circa $2500 msrp ...

not so clearly different. 24/96 to Redbook, in most cases,

very very hard to find an audible difference with what I can discern

are 'good' recordings. Sort of like: "I think I can!!" ;-)"

 

You started way too high. The differences between the DACs built into quite average digital sources and expensive standalone DACs are often small. The differences between really well-designed stand-alone DACs in the low $100s and multi-thousand dollar DACs are often minute. Given enough power and a great recording and master, good audio is 95% transducers. We're sweating the small stuff, which is OK, it's what we do.

 

Tim

 

[Daphne]

After reading all the comments on this discussion, I returned to the Dan Lavry interview to read for a second time. Lavry did not give anything away. His answers were vague at best, less that one can find in a basic primer on the subject of DACs.

 

However, visit the Lavry Engineering web site and you will notice that Dan Lavry designs some very interesting AD and DA converters. Allow me to change interesting to complex and sophisticated. Good god, he actually installs small ovens to maintain heat on temperature-sensitive components. His Audiophile DAC takes ten minutes to warm up and self-calibrate.

 

I have read about a number of DACs on this forum, but nothing like Lavry Engineering is producing.

 

Lavry was most vague when asked what is the secret to a good sounding converter. "Engineering is the art of optimizing and compromising. Making gear musical does take lot of experience and understanding of the ear. I never underestimate the ability of a good ear."

 

When converting digital back to analog everyone has a philosophy. They optimize the signal before the converter chip or after, or both. Each solution has its compromises. So which solution has the least compromise regarding sound quality? It seems that subject is a matter of opinion.

 

I like the way he reduced Psychoacoustics to "understanding of the ear." Now that I think about the interview, it read like Lavry was speaking to a classroom of preteens.

 

Bob Katz offered up more information in his interview, but not by much. Recording and mastering engineers hold their secrets close, their knowledge, experience and expertise is how they earn a living. Limiting their competition is good for business.

 

I found Bob Katz's comments rather interesting on the future of digital audio heading to surround sound. I have several Audio DVDs with 5.1 surround, and the mastering is done like a movie soundtrack. The listening experience is not very pleasant.

 

I heard a Revel speaker demo in surround sound which sounded fantastic, but the source was a demo 5.1 SACD. Until that point, I was unaware SACD was even available in surround sound.

 

daphne

 

 

 

 

[mpmct]

So ... all of you with golden ears are no doubt

willing to submit to a test in its parameters roughly equal to this double-blind? ...

 

http://www.bruce.coppola.name/audio/Amp_Sound.pdf

 

If you lose, a new BMW 135i, racing red, delivered to the challenger?

( Only about $36k USD -- in today's dollars of course. Tomorrow, who knows? )

 

Dare-est any of you? ;-)

 

( Relax, I'm just having fun with all of you.

It's only stereo fer godsakes! :-) )

 

 

 

 

 

 

 

 

[The Computer Audiophile]

I like the good sense of humor mpmct! The high-end world can use as much humor as it can get!

 

[tfarney]

I'd love to see a similar test run with DACs, starting with something as humble as the $200 Beresford, working up to something in the stratospheric high end. Very different things, amps and sources, but I'd still love to see the testing.

 

Tim

 

[tom]

I feel fortunate to have good ears.

 

The difference between 4416 and 9624 to me is clear as day.

 

There is nothing I can tell people other than "it is clearer" when I describe 24 bit tracks.

 

I will say this though. I find the difference between 9624 and 192 24 to be extremely minor. With an upgrade in equipment, maybe I will hear a difference, but at this time any difference in sound quality is negligible. The only album I have in both these qualities is Hotel California.

 

I do have many albums in 9624 and listen to them all the time. I went through the entire list on dvd-a.info and purchased a whole lot. My main problem with being an audiophile is the lack of contemporary music that is at all close to mainstream. The best newish album I have is Paramore Riot at 4824.

 

I find Dan's comment about sample rates in the article/link to be VERY interesting.

 

 

[Ashley James]

Tom

 

This particular Album is surely an old analogue recording and the equivalent of an 11 Bit Digital one! I'm not sure that re-mastering to a higher Bit rate could improve the sound.

 

Ash

 

[cs]

Earlier in this thread, I noted that not many people can hear much above 16kHz, and I certainly can't.

 

It would therefore be very interesting to make a recording at 44.1kHz sample rate, but with the audio bandlimited to 16kHz to begin with. This would allow a very benign filter characteristic to be used, both in the recording ADC and in the replay DAC. I suspect that 16 bits would be quite sufficient too, although it would make sense to use 24 bits for the initial signal capture and editing.

 

As a related aside, it is also worth noting that VHF FM stereo transmissions (in the UK at least) use digital links between the studios and the transmitters, which operate at 32kHz sample rate and 13 bit resolution. Provided you have a good tuner and a fairly strong signal, it can sound really quite good.

 

Chris.

 

[PeterSt]

Why is it, that *nowhere* on the Net descriptions are given like mine below ...

May it be so that nobody is dealing with the actual matter, or is it me just flawing hehe.

 

I am really sorry for this super long post. It just happened ... :-)

 

For those who don't know, background : The software I write (XXHighend) in order to let virtually any DAC perform optimally, is not much different from the software that drives the DAC itself. The difference is, that the way I do it (software wise !) is taking the physics of all for granted, and squeeze out the best of *that*. One of the physics is the WAV data file ...

 

Now, first funny thing is that we all tend to judge on what we (not) hear and perceive, which creates the different standpoints because we all listen subjectively. This is funny, because we also could start off at the base of the matter, and try to explain how it would be possible that we e.g. *not* hear a difference, while we should !

 

As a sidenote (which possibly should be the main topic) : I did not read the article about the optimal sample rate, because I already know it is true by my own findings. *But*, this is most probably for different reasons than Dan Lavry comed up with (if at all), and which I wrote about (3 months ago or so)in a post meant for this forum, but which became too long so I didn't post it.

 

On to the matters, now first my conclusion, which is not subjective an not heard through a system which coincidentally can show it. No, it is heard through ears which know what to watch for, and through ears which have the reference. And of course the knowledge of what it actually going on. Yes, a precedent.

Of course you can hear the difference !! The diferences are all over.

 

So, can I explain these differences to you ? nah, I don't think so. And to give two example of this impossibility, you'll see how difficult the discussion is when the discussion starts at the end (the listening) :

 

Example 1

 

A 96K recording can sound less precise and sometimes more harsh opposed to 44K1 because one of the components can't follow while the others can.

 

Think analogue here. First the software does all it can to control the DAC in such a way that it is the most precise. Say, it's more sharp, more to the point (unbelievers, just bail out now :-).

We got ourselves the fastest amp we could find, so it will be able to follow those actual digitial steps which *are* in the base. And yes, we got ourselves a non oversampling DAC to pertain the original waves better, which are ... digital here. Oops.

We also got ourselves a TVC, which is as passive as possible, but which also is known to be prone to have a non lineair impedance response (like say, passive xover filters).

What may happen now, is that the superbly followed "square" e.g. synth waves of the digital stream create harmonics which are allowed by themselves, but the TVC chokes in. Next, our super fast amp is able to show that mess perfectly ...

 

The moral of this example : When the source wouldn't be as precise (so start with an oversampling DAC for that :-) or the amp wouldn't be as fast ... either "solution" wouldn't show the anomalies the 96K file incurs for. And hey, this is just one example from one angle, of which there are so so many, and they all contribute to the result. Bad, worse, better.

So is 96K worse than e.g. 44K1 ? of course not. Does it show worse ? yes, with this example it does.

 

Example 2

 

I have a 44K1 file and a 192K file. It may well be so that the 192K file does not sound better. Why ?

 

Now, anyone not taking into account at least a few of the below reaons won't be able to know what to actually look for, or just judges wrongly based on pure coincidende ... (and the below ar just a few reasons, there are more) :

 

When my software is influencing the DAC as best as it can, it needs the lowest latency possible. With the equipment I use, this means 48 samples (around 1 ms) for a 44K1 file. However, a 192K file requires 128 samples. Nothing to do about that, it is (my) hardware impeeded.

The 192K file will always sound *worse* because of this reason alone. Net it can sound better though (see below to the point description).

 

With both the 44K1 and the 192K we may wonder where it came from. Was it a 192K native recording ? then the 192K file has the advantage. The 44K1 looses already because of necessary dithering.

Was the original recording in DXD hence 352K800, then the 192K file looses because of the same reason (and note that 352800 / 8 = exactly 44100).

Again, net the 192K may sound better, but remember that there is some "base" now which is just wrong, while the 44K1 hasn't got that wrongness.

 

Since I am a firm believer of non oversampling, just to pertain the squares we'd loose at heavy oversampling (and besides synth sounds there are more squares than you'd imagine -> think of rolling surfs, or thunder for that matter), my 96K nos DAC just exists, while no 192K nos-DAC *can* exist. They are all sigma delta, so need internal oversampling just because of the principle.

For this reason alone, a 192K file will *always* loose. Net too. Ok, it depends on the type of music a little, but thinking of complex harmonics (which emerge from squares !), it's actually all over.

 

The moral of this, is that we are just not able to judge without incorporating the above, the more things in this area we can come up with, *and* the combinations of them.

 

The to the point description

 

Yes, time for that.

With the above stuff in mind, there is this very firm base in the digital file itself. This is not some vague thing of which we might wonder whether we are able to hear it, no, this is the other way around. If we don't hear it, we must seriously wonder what is wrong with our system. Remember, our system as a whole ... see above examples. Here goes :

 

Our real life consists of transients. I am not sure whether somebody in this world took the trouble of what transients can be per instruments, women voices, cymbals, etc., but they are there, they are a given fact, and when not taken into account properly, things are unreal.

Now I hear you thinking ... so a 192K file is of course capable of presenting steeper transients ? ... but no :

 

A 44K1 file is not able to present the slower slope of a transient. They are all too steep !

 

How come ? This is very very easy to understand, but first you must know about the data in that 44K1 WAV file :

 

You can just see that a common WAV file has transients of way over half of the voltage range. Thus, say the DAC goes from -2 to +2 volts, and the 44K1 WAV file shows voltage jumps from e.g. -1.4 to + 1.6 in one step. Even the near complete range takes place in one step occasionally.

Sidenote : XXHighEnd trips on "crack detect" for safety reasons, and if more than 25 steps of over 80% of the total range are detected, it trips. 70% already won't do, because each album would trip on it. I started off with this at 0.1% because I thought steps really wouldn't be larger, but then millions of "cracks" are detected ... and I was startled.

 

Now think of what this actually means : We have two subsequent samples, and one implies a voltage of -1.4 and the next implies +1.6. This is a total jump of 2.8 Volts from one sample to the other, and it already pesters the slew rate of the DAC. In effect this implies a dB (SPL) jump of 70% of the total volume you are running at at that moment - whatever this is - but within 1/44100 = 0.0000226 seconds.

 

In between lines : think about above examples again, and what would be able - and what would not be able to follow.

 

For easiness, we compare this with a 176K400 file. Theoretically speaking, the voltage jump will be there within the same amount of time. However, it will be spread over 4 individual steps now. Theoretically more smooth, and again theoretically the DAC will be better "controlled". But careful, because the total slew rate needed doesn't change, because all keeps on happening within the same time span.

 

Ok, still with me ? then back to the nos DAC and the squares. But keep in mind the latter.

 

When a square is fed to the DAC, an oversampling DAC will round that square. You could say this is related to pre and post ringing, and when the square is heavily oversampled, a sine will be the result. Ha, so now all the equipment can follow. Yeah maybe, but it is not the truth what you listen to.

Looking at the far too steep (n fact straight) slopes, if something is acting as a square, it is right that. Now, the non oversampling DAC will put that right through. But keep in mind, this "square" at the 44K100 file is just too much of it. IOW, it would be more roundish when the same data would be in a 176400 file. 4 times more roundish.

The characteristic of a square is that it produces harmonics, including more total energy compared to the pure sine of same amplitute (which is logic, because the amount of time the amplitute of a square is at full plus and full minus, is longer).

It is hopefully easy to see that an amplitude which rises too fast (and drops too fast) will stay up (and down) longer. It is more a squeare than intended, and harmonics hence distortion of a too high amplitute are the result. Make this a 176K400 and you're 4 times better off. Make it a 352800 and it's 8 times better.

 

Fake a 352800 because of oversampling and this is not necessarily better for the net result.

 

Now what ? So yes, the harmonic distortion because of inappropriate squares will be less because there are less inappropriate squares. But now think further, what we actually did with oversampling :

 

With the earlier example of the trancient covering 70% of the total range, we could map this to decimal figures for easy thinking. Thus, with 44K100 we have 65536 possible decimal values, and 70% is covered by the one trancient jump. This means 45875 possible decimal values, and now we are going to upsample this by a factor of 2. So what shall we do ? divide 45875 by 2, and add the additional step right in the middle (if possible) ?

So tell me, how big is the chance that would be right ? yep, this is a chance of one out of 45875, so this will *not* succeed.

 

The only thing upsampling may do, is eliminating partially (!) the harmonic distortion caused by something which is wrong at the first place (the too steep transient, hence the too low sample rate used at recording), and as a side effect we will have faked reality in a way our brains can't cope with.

Or let me put it differently : with all the effort I took myself in a decent upsampling (with or without AA filter) the result is *always* that the native 44K1 sounds the best. Net.

So is this the definite conclusion ? nope; remember the example about the buffer of 48 samples which can be used with th 44K1 file. This is 96 for 88K2 and therewith already worse for that part. So I just can't tell really.

 

 

Somewhere Chris C. asked about the video upsampling. IMO a complete different matter;

The pixes to interpolate - in fact not much different to our audio samples - don't require much more than anti aliasing and the mathematical lineair gradients needed between the real pixels in the file. What can go wrong with it ? It's all in a 2D plane, and as long as all is not based upon too much mosquito distortion (coming from too heavy JPEG compression), this just works out for the better. Undoubtedly, and everybody can just see it. But the point is, this is not fragile.

In order to understand this latter please listen to this small story :

 

At some stage during XX development strange things started to happen. Btw, those who know me, know that I have some weird theories about standing waves, and this is kind of related to that.

Ok, think of things getting so "sharp" in the positive sense (pinpointed at micro level), that collisions in mid air happen. Right, collisions happen all the time, but when things are sharp enough, they interact in a positive way. Just think of the violin and how it produces her complex harmonics. This is not the string, not the cabinet, not the kind of wood used, it is all together which means that things are mixed. They collide and produce the total sound because of that. Hold a microphone at the proper distance, and you'll be capturing all.

For comparison, now think of small microphones at the smaller distance, each capturing some unrealistic individual sound of the instrument, and play that back through loudspeakers ...

 

And now you think I will say that this can't work. Haha, it just can ! the harmonics will just be recreated in mid air, and there is no reason why they would not ! (of course the violin is a stupid example, but you'll get the idea). This means that not captured audio data just can be recovered by the physics of air.

 

The more "sharp" hence pinpointed the system as a whole can output the waves, the better they will interact, and while I have proven that this happens, I never could prove that it happens wrongly.

For fun : this also has downsides, because when waves interact at this (micro) level, they also phase out eachoter. Result : plain distortion (because it happens at a very high frequency). I could personally solve this though, and never mind how :-)

 

Allright, the moral of this latter little story, is that these things all don't - and can't happen with video. The only thing which can happen there IMO is wrong AA appliances, with the result of moire or banding and the like. Easy stuff, although it requires much more processing power (just much more data to process).

 

 

Concluded

 

The comparison between e.g. 44K1 and 192K is not about the frequency domain. It is about the amplitudes in the time domain incurred by the sample rate, that causing unreal or more realistic transients, with the further effect of when them being steeper, the less chance the equipment can follow. All together there is much much more to it, and all plays around harmonics, those by itself unrealistically created by squares which shouldn't be squares, or them unrealistically not being present because the actually present squares where fainted by means of oversampling.

When you know about this and have heard what causes which, you keep on hearing it. This is your reference.

When you jump right in the middle of all this, all is one big pile of apples and oranges, and what you like with one recording, doesn't work with the other, and thus it is ... *not* that recording. It *is* about all the prerequisites which have to be build up from the ground, and which takes virtually ages because each element has to be covered for (first understand, then tweak).

 

All is a complex of the too square data in the digital file on one side, and the very precise equipment just showing that on the other. We should be lucky that loudspeaker drivers can't follow that, which on the other hand will create harmonic distortion by itself.

This is *not* about the smallest step in the digital file which is too rough (22050 Hz is an exact square at 44K1) but it is about the impossibility to have more steps sideways (time domain hence sample rate) where these more steps the most obviously do exist (I don't think nature and physics allow a transient of, say, 70dB to happen straight up).

 

Thanks,

Peter

 

[tom]

Ash, I'm sure it can. I don't know what you are trying to say. I can hear the difference between 4416 and 4824 for the Paramore. And there are no albums available at 192 24 that are modern (2000s) as far as I know.

 

[Mallette]

Need to visit more often! I am a long time and constant denizen of the Klipsch Fora, mainly the 2 channel Forum and don't get out much. In that august group I am the voice of one crying in the wilderness about the value of digital. However, I won't get into that but get on with this topic.

 

What I read here jives with my own decade-long experiments that started off trying to determine why CD's just did not often satisfy like a fine LP. Again, that is a long story more suitable for another time.

 

To get to the point, I've recently settled on DSD as the quite possibly the "universal" high res format and would like your expert opinions. My reasoning is based on listening tests. I recently acquire a Korg that supports DSD and have made a few test recordings, all from LP as I've had no opportunity to record live music with it yet. The most telling test was made from the famous Crystal Clear direct to disc Virgil Fox record. If you've heard it, you know it is about as good as it can get in the vinyl world. With a couple of well-trained ears present I did a sync between the LP and the DSD recording of it and performed a blind A/B. No one could tell the difference. I will repeat this Saturday with a larger group of golden ears to see how it fares. But wait, there's more...

 

I have transcoded this and other recordings from DSD to several PCM bitrates and am finding no degradation that I can hear. I've always been able to hear PCM downsamples from uneven multiples but in this case I am not hearing any change at all except for that which one would expect going to those PCM rates that cannot contain as much info as the DSD. The higer (24/88.2 or above) rates sound identical.

 

So, I am thinking that DSD is the state of the art format for digital recording in that it can allow for transparent transcoding to all PCM sample rates.

 

Your thoughts?

 

Regards, and congratulations to CA on your growth!

Dave

 

[The Computer Audiophile]

Hi Tom - David Gray has a 24/192 DVD-Audio disc available. I have it but can't think of the title.

 

[The Computer Audiophile]

Hi Dave - Thanks for the kind words about the site. There are some pretty cool developments in DSD and DXD availability for music servers coming fairly soon. I'll have some news in the next few days about some incredible DXD material.

 

[Mallette]

In 2003(!) I made a 24/192X4 channel live recording of the Asylum Street Spankers with one of my own PC digital recording rigs. The files are enormous, as you can imagine. However, the results, when played on 4 identical speakers equidistant from the listener, are downright spooky.

 

Dave

 

[PeterSt]

Referring to my before post (too long to read anyway :-) here's my 2c from the other angle : listening to te result :

 

Because of the other parameters we just have to deal with (see other post) it is very rare that even a 96K file sounds better than a 44K1 file. The only real example I have myself of 96K really being better, is Brothers in Arms 20th anniversary edition. But that's about it.

That a random 44K1 sounds very different from a random 96K is for me without doubt though.

 

It is not easy at all to determine what is going on in all parts of the audio chain, but - being busy with working on improvements on loudspeaker systems, amps and DACs - it more and more happens that we come to the "weak" conclusion that we are going too far with things. Weak, because we can't think of another reason why "a" best solution doesn't sound better at all.

 

On a sidenote : I am talking about a complete chain of products i.e. player, DAC, preamp (or no preamp), amp, speakers, which all are developed by persons in my own environment, and those persons working together to get the best chain possible. The result in my room is a complete natural reproduction of the instruments you can think of, yes, including cymbals.

But since the natural reproduction of an instrument is different from the natural reproduction of a performance (stage etc.), there's always things to improve on ...

 

Allow me to give an example :

We have a DAC, and it works like all other DACs when controlled via a PC.

We have the very same DAC, but in a version the PC just cannot influence at all. It's just doing its own job, and all theories say it should be better.

@Gordon : you know what I mean. And no, it is not your DAC I'm talking about.

But, although many people rave about it, for me, with my own consistent setup, the version with "PC influence" sounds better. What ? far better.

The version without influence is virtually jitter free (ok, nothing is free of jitter really), and it looks like all is coming through "too pinpointed". Think at the micro level, and try to imagine those very tiny variations in amplitude which are just passed through more 1:1.

 

Now, with the most fast amp we could create, and the most sensitive speaker (what about 115dB), this is just passed through the best.

We often talk about solutions like moving in a slower amp again. But hey, what is that for a solution ...

 

Although I can't put it in proper wording, and certainly can't justify it all, with my knowledge of things it looks like a 96K file (opposed to 44K1) just makes things worse. IOW, even more things in the chain can't cope (I know, this is opposite to the transient story from my before post, but just so many things are going on).

To name another thing I never hear of, but what for me may make differences, is the speed of variation the higher sample rate file will incur for. Thus, a frequency of, say, 18K for a loudspeaker driver is one thing, but the variation in amplitude of "the wave file" of 96K in theory, is quite another. For me this can well mean that the driver must be capable of 96K variations in amplitude, which it possibly can't when the specs say it's capable of 30KHz (don't think analogue here which will incur for infinit amplitude changes -> that's harmless "sine" analogue !).

Opposed to this we have the 44K1 file which will incur for only 44K1 of those movements in the driver, but, now these individual movements are larger. So here we are again : apples and oranges.

 

I know, I am moving at levels possibly not suited for me (no signal processing in my education :-) but one thing you can trust me for : seeking for explanations of phenomena which are not explained so far.

 

Last example to hopefully make clear what this is all about :

 

Everybody will agree that loosing the preamp would be better than having it in. That is, if you have arranged for the impedance problems that may come from that). Thus, passive or not, we moved out a device in the signal path.

Many of you will already know the enormeous difference in sound when the preamp is left out. And, many will agree upon the sound being better without.

But now you may wonder what actually happens, and for sure those who came to the conclusion that without preamp it does *not* sound better;

 

Your preamp will 100% sure round those squary signals it is fed with. And, the better the DAC passes those on, the more the preamp will be "actively" rounding those off. Now, leave the preamp out, and those nasty waves will be passed on to the main amps. Then, when the amp is fast enough and it doesn't show much ringing (pre/post) it will go right into the speaker.

 

Is that good ? In theory no. So, is it better to round everything, e.g. by means of a preamp, filtering cables, etc. ? hell no ! I mean, in that case we really don't know what will happen with our precious data.

 

All 'n all I managed with all this (and without preamp), but it really took a year or so of looking at the data, analyzing, understanding and interpreting impedance effects.

 

The 1:1 principle (square in, square out) is what I still hunt for. With 96K it possibly works counteractive, and with 176K400 and up it just can't be done (oversampling DACs only), *unless* I'd be ready to use a 16bit DAC for 24bit files ...

 

Peter

 

 

[Ashley James]

Surely this is an old analogue recording?

 

Ash

 

[DavidJPettifor]

Congratulations on your 100th post on the CA forums, Ashley! Do I claim my ADM9.1s now for spotting this?!

 

--

djp

 

[PeterSt]

The back says : CD: High resolution Digital remastering by Keith Blake.

(btw, I didn't compare because I was kind of fed up with the original, but I'm sure the tracks are extended and (sometimes very) different).

 

The cover of the original (1985) says : "A full digital recording". So ...

 

Anyway, indeed this is the only hirez album which was *really* joyful to listen to.

 

A remark of a different kind :

I think we all know never to judge a system with Dire Straits, because it just sounds good always.

Ok, this might be true within itself, but wait until you hear Dire Straits when everything in the chain is really good ! There's more in there (detail etc.) than at least I expected for decades, and now it looks like this hirez album shows the same ?? It just *is* better than anything else (I got, which is 27 albums only).

 

Judging hirez albums is not easy at all, because of the repertoire which is relatively small.

Also, proper judging for me means that cymbals and drums must be there, while downloadable promo's (like from Linn) are too "classical" for that. Btw, they have genuin 192K and even 352K800 (DXD). The latter only plays downsampled to 176400 for the normal human being without 384K DAC :-).

 

(Might you be interested in playing such a 352K800 file, I made XXHighEnd explicitly (for Linn) to do that properly, but could listen myself to 176400 only (lacking DAC). Note though that usually two subsequent of these large tracks won't play in XX, because it plays from memory, and the files are monstrously large).

 

And FYI : The McIntosh Demonstration Reference Disc contains 15 192K tracks.

Peter