[audiobat]

No generalization, how about an example.
You have always cooked using a microwave, you cooked all your food you entire life in that microwave.
So your lifestyle is suited to that process, it's quick, it's easy, and it will in fact cook anything you want to eat.
Forgive me everyone who loves a great steak, but you could even toss in a fantastic $50 filet mignon (like one from Allen Brothers!) into that microwave and it will cook it, (sort of).
However it is never going to be an Allen Brothers filet mignon cooked on a wood fired grill, will it?

I understand that you have no TT or records but I will bet that you are "thinking" about something new for your system, right? It will cost money whatever it is, Cables, preamp, speakers, who knows. I say spend that money on a TT, phono preamp and LP's, you will find that you have a much better sounding system than you thought.     

   

[Wayner]

What makes this such a moot point is how many of you are going to actually purchase your whole music collection again as high res downloads?  I've got almost 10,000 CD's.  It's not happening here.

There's so many issues that aren't even being discussed, it's just another topic that makes everyone crabby.  I've had the opportunity to listen to enough digital, analog, high res etc, that I can be very happy with whatever format it's released on.

I have many recordings, regardless of format, that I consider "reference" recordings. The 2 CDs I mentioned earlier are on the list, but there are several vinyl entries that are as spectacular as their CD cousins.

I have long held the notion that it is not the medium used, but rather the recording itself, that separates the awesome from the ordinary.

Wayner

[trebejo]

Going back to the OP, the idea was to talk about the science of all this.

Here's another scenario: your buddy tells you that he spent last night listening to music on some kind of recorded media and it sounded great etc. and he invites you to come by and listen. Now if you have some good sound science that tells you that his recorded media will never be as accurate as some other format, what do you do, tell the dude that it's not worth it to go there? No, that would make you a bad buddy! So you take some stuff to consume that goes along with the music and enjoy it and have a fun time.

I guess sometimes when we get on these threads here and tell people that their CDs or their LPs or their flac files are not accurate enough, it sours up fast because it's not a positive vibe to introduce into the listening experience, man...

BUT... the OP was not doing that! He was openly sticking to the science!

Sooooo... if you're not bringing science into it, then you're the one raining on your fellow audiophile's parade. Said audiophile wanted to talk about the science of the thing.

There is a bit of an antiscience wave going 'round lately, usually voiced by people that, well, how to put it charitably, they didn't exactly ace their way through math and science class in high school...

btw Shannon's theorem is obviously correct and not to be questioned, since that process took place before it got to be called a theorem. So as far as upsampling is concerned, it's mostly a crock for listening although there may be some cute tricks for artificially enhancing the music during the production process that are available with higher frequency sampling (I wouldn't know about that). But for playback, 2N sampling rate yields resolution up to 1N frequency and that's just basic fourier theory.

The other aspect to this is the higher bit count leading to greater sound level resolution. If the noise floor on our audio devices is low enough, this can make a difference, but does that EVER happen? Do our homes have a 60-100db noise floor? None of mine ever have.

I went to the opera the other night and took the SPL meter along. The soprano on her solo arias was around 75 db (back rows of the Chandler). On a previous occasion I took the meter to the Disney, mid-hall (in x-y-z axes) and Mahler's 6th reached peaks in the low 90s and typically played around 75db. That tells me that in terms of loudness, I can probably get the same fidelity at home with 16 bits as I can get sitting at the Disney (as far as noise floors, etc.).

If the very same recording sounds quieter/deeper/etc. with 24 bits than with 16 bits then either your house has a noise floor ca. 100db or the people producing the 16 bit version did not get all they could out of the technology. With this answer I don't blow off what someone is certain that they have heard, and I don't blow off the hard work of math and science that has taken centuries to accumulate and, for some, mere seconds to blow off as inconvenient.

[Freo-1]

Going back to the OP, the idea was to talk about the science of all this.


The other aspect to this is the higher bit count leading to greater sound level resolution. If the noise floor on our audio devices is low enough, this can make a difference, but does that EVER happen? Do our homes have a 60-100db noise floor? None of mine ever have.



If the very same recording sounds quieter/deeper/etc. with 24 bits than with 16 bits then either your house has a noise floor ca. 100db or the people producing the 16 bit version did not get all they could out of the technology. With this answer I don't blow off what someone is certain that they have heard, and I don't blow off the hard work of math and science that has taken centuries to accumulate and, for some, mere seconds to blow off as inconvenient.

There is more to this than just the noise floor.  There is also the smoothing of the stepped digital output waveform back to analog sine waves. 

The report presented is flawed, and attempts to validate a pre-conceived argument, period.

Science is a easy word to throw around, but is it proper science, or junk science, or something in-betwween?

As adroitly pointed out by Newzooreview eariler, there are issues with this.  The author is no doubt well versed in audio, BUT, that does not mean he has got it all right. 

The only real way to resolve this is if the AES got serious and commissioned a panel of experts to study the subject with a proper set of entering arguments, which spell out the process, procedures, and methodologies to sort out fact from fiction.  It's not going to happen, so we are left to debate this.

I would further venture the vast majority of audiophiles who have both hi res and redbook CD recordings will testify that "in general" the hi-res version more often than not sounds better.  I seriously doubt that is in question.

[trebejo]

There is more to this than just the noise floor.  There is also the smoothing of the stepped digital output waveform back to analog sine waves. 

The report presented is flawed, and attempts to validate a pre-conceived argument, period.

Science is a easy word to throw around, but is it proper science, or junk science, or something in-betwween?

I don't get it, the frequency information is preserved as per Shannon, right? Science, Shannon, Fourier, trig series... there's lots of words to throw around, and big books to grab them from.

[Wayner]

Well, Frank's very first post has the article that says "it is in question".

A crappy master will then sound better cause it's in a hirez format?????

Wayner

[Freo-1]

I don't get it, the frequency information is preserved as per Shannon, right? Science, Shannon, Fourier, trig series... there's lots of words to throw around, and big books to grab them from.

No?  Forgive me, but I think you get it.  For more details (and variance)

http://www.hifi-writer.com/he/dvdaudio/dvdvscd.htm

http://en.wikipedia.org/wiki/Super_Audio_CD


http://www.analog.com/static/imported-files/white_papers/ADI_SHARC_2148x_and_2147x_HD_Audio_Paper.pdf


http://www.itwriting.com/blog/articles/is-high-resolution-audio-like-sacd-audibly-better-than-than-cd

[trebejo]

Come on, don't just list a bunch of links, explain what you mean in a sentence or two, I'm sure you can do it.

[Freo-1]

From the DVD Audio wikipedia:

"Sound quality
 
From a purely technical standpoint, the audio resolution of a DVD-Audio disc can be substantially higher than standard red book CD audio. DVD-Audio supports bit depths up to 24-bit and sample rates up to 192 kHz, while CD audio is 16-bit, 44.1 kHz. In both cases, the source recording may have been made at a much higher bit and sample rate, and down-converted for commercial release."


"  Compared to the Compact Disc, the much higher capacity DVD format enables the inclusion of either:
 Considerably more music (with respect to total running time and quantity of songs) or
 Far higher audio quality, reflected by higher linear sampling rates and higher bit-per-sample resolution, and/or Additional channels for spatial sound reproduction."
 




This speaks to what Wayner was referring to on recordings:

Many DVD-Audio releases are older, standard-definition audio recordings that have been remixed in 5.1 and upsampled to DVD-Audio's higher resolution. However, the fidelity of the upsampled audio will be limited by the source material quality, when the master recording is in digital format, and may not exceed the quality of existing CD releases of the same albums. Recordings that are made using high-resolution PCM encoding can be released with a resolution that is higher than standard CD.


So, obviously, there is better resoultion with both dynamic range and frequeny extension with high resolution.

[Wayner]

But the regular old CD could produce the same audio if it were not restricted by compression.

I think we have 2 issues going on here. The CD was made for the masses and treated as such. The hirez was created for audiophiles. However, that really means that the intended purpose and marketed group steered the treatment of the musical content. So, any kind of comparison between the two may be flawed.

That may not be a flaw of the medium, but rather the application.

Wayner

[trebejo]

Yes, of course, and with a 192 khz sampling rate you will be able to retain the fidelity of air waves oscillating up to about 95 khz. That includes about two octaves above the hearing of the acutest pair of 10-year-old ears around.

So in the frequency realm, this "gain" is utterly moot.

In the amplitude realm, you go from about 100 db to 100+I forgot how many db and then that brings us back to what I said before.

I know this circle, we all know this circle, we only keep running around it because people want to ignore the math and science.

[mfsoa]

I don't understand how a high frequency (like 10 khz) sine wave can be perfectly reproduced if there are only 3 or 4 samples per cycle. I know it can be produced, but can it really be reproduced to look like an actual sine wave?

As far as bit depth goes, I've heard (and just taking a guess at the actual bit values here) that since you absolutely do not want to exceed 16 bits of data on the very loudest passages, you might actually be using say 14-15 bits at most for these very loudest passages. So if a full blown symphony orchestra uses 14 bits, how much does that solo flute use? Or that triange ding waaaayyy in the background?  6 bits? 8 bits? Less? And the bits available to the overtones or reverb of that solo flute or triangle? Very few, I'm guessing.

So my (probably flawed) understanding is that just because you can create a 10khz signal at very low level using 16/44 sampling, you might only really be using 4-6 bits, with 3-4 samples per cycle.

This is the reason why cymbal decay can sound so poor on CD - as it decays, you lose resolution (the ability to accurately describe the event) and the sound ceases to be "cymbal" and becomes a non-descript hiss. (amateur drummer here so sensitive to this phenomenon)

For this reason, the same signal can be much more accurately recorded/stored/recreated if using higher bit rate and bit depth techniques.

I think.

[Freo-1]

Yes, of course, and with a 192 khz sampling rate you will be able to retain the fidelity of air waves oscillating up to about 95 khz. That includes about two octaves above the hearing of the acutest pair of 10-year-old ears around.

So in the frequency realm, this "gain" is utterly moot.

In the amplitude realm, you go from about 100 db to 100+I forgot how many db and then that brings us back to what I said before.

I know this circle, we all know this circle, we only keep running around it because people want to ignore the math and science.

No, you chosse to accept an incomplete and flawed test as the gospel. 

Again, I point to Newzooreviews adroit comments regarding the shortfallls on the approch and controls. 

Engineering 101 teaches us to challenge plans, approaches, assumtions, etc. to ensue adequate controls are in place. That did not haappen here (with this referenced report).

The difference with SACD/DVD Audio sound is as opposed to CD is REAL.  The vast majority of audiophiles readily hear it. 

Wayner's points about mastering also need to be factored into this discussion.  Trying to prove CD can sound as good as hi res is difficult to nail down with certainty.  To be sure, there are great CD masters, and sub-optimum hi-res mastered recordings out there. 

[trebejo]

Just a couple of very brief comments (otherwise I have to start totally making stuff up).

The number of bits is, roughly speaking, a volume thing. 16 bits gets you 2^16 levels, so you take the log based 10 of 2^16, and then you multiply by 20 and that gives you the decibel range. So I get 96 db with that calculation. Then one loses a bit here and there for whatever reason. If you lose a bit, you lose... 6db. So if you are worried about only using 4-6 bits, you have lost 60db of signal and then yes there is noticeable degradation.

The sampling rate then gives the number of strings per second. So 44 khz gives you N strings per second (shall we say 44000? It's something like that). 192 khz gives you 192/44 = 4.36 N strings. So the size of the new music file should be about 4.36 times as large as the previous file IF you stick to 16 bits. If you do 24 bits, then the new file should be about 4.36*(24/16) = 6.55 times as big. So if the original CD had 500 MB then the new CD should have about 3.3 GB, roughly speaking.

As to why Shannon's theorem works, well, Fourier theory did cause Msr. Fourier some grief during his lifetime, when the French academies were outraged at the notion that one could deconstruct a continuous function as the sum of cosine or sine functions. Some fundamental concerns remain as far as our understanding of various cardinalities of infinities, whether sets can be well-ordered, etc. so in principle one could remain skeptical about the merits of Fourier theory. In other words, if this all sounds fishy to you, well, you're not the first and you are in some pretty respectable company. However, in practice, it is used all the time (e.g. this computer works because of it) and it works extremely well so as far as I know nowadays it is totally accepted and not the least bit controversial.

The perhaps more amazing thing is that a discontinuous function can also be given as a countable sum of sine functions or cosine functions. 

[trebejo]

No, you chosse to accept an incomplete and flawed test as the gospel. 

Is this where I shrug my shoulders and say "whatever"? The process that you are describing as incomplete and flawed is written up in a zillion textbooks but if you can point to the one where they say that Shannon messed it up then go for it. It's just trig series...

Wayner's points about mastering also need to be factored into this discussion.  Trying to prove CD can sound as good as hi res is difficult to nail down with certainty.  To be sure, there are great CD masters, and sub-optimum hi-res mastered recordings out there.

A clear issue with mastering and higher sampling rates is that when a lossy filter is applied, if you have more bits than you need then when you are done you still have enough bits and so things work out better.

The more famous example of the above is when you have, say, 32 bits to work with, you multiply by 1024 so now you have essentially chopped off the last 10 bits, then you divide by 1024 and you cannot get back the original number. It's hysteresis for binaries.

[coverto]

A key point being neglected in this whole discussion is the fact that, however superior the data in a hi-rez file may be to that of a redbook file, the ability of our systems to accurately process that hi-rez data and render it into music is seriously limited at this juncture.

Indeed, here's an interesting piece on the subject by a manufacturer of digital audio equipment who is quite emphatic about the limits of digital in this regard:

http://www.coreaudiotechnology.com/blog/?p=57#more-57

Among other things, he makes the point that accurate 24-bit reproduction is impossible for even the best DAC chips that are currently on the market. That's before you even consider all of the other pitfalls that create amplitude distortion in a digital stream - heat, power supply noise, RFI/EMI, etc. These issues keep the practical ceiling closer to 20 bits, he says, even with the best tested equipment. That's far out of the reach of guys like us, whose systems typically are doing well to manage 12 or 14 bits, much less 16 bits.

I'm no expert here, but this fellow has some credibility with me, as his products are currently producing the best digital I've ever heard in my system, or probably anywhere else for that matter. That said, I must admit I've also heard improvements in some hi-rez files over his gear. I am guessing this is mainly due to superior mastering for hi-rez files, although he acknowledges these audible improvements and floats a few other hypotheses that attribute them to "pleasing distortions" in the digital signal, or the fact that hi-rez data is cut into smaller slices, allowing distortions to do less damage to the overall signal.

Indeed, there is much with hi-rez that remains a mystery to some very knowledgeable folks out there. And of course, it must be emphasized that what sounds good to the ears, sounds good to the ears, and the reason why is of secondary importance. Nevertheless, I am persuaded that the true capabilities of hi-rez digital technology currently are more limited than many people think. 

[rbbert]

The usual question here is not whether all 24 bits can be reproduced correctly, but rather whether the greater amount of data in the 20 or so most significant bits that CAN be reproduced correctly results in better sound than 16 bits.

[coverto]

And likewise, there's a question whether those extra bits that CAN'T be reproduced correctly are mucking up the sound such that it sounds worse than 16 bits. In my experience, it can cut both ways. Accordingly, I've found myself growing pretty careful and my purchases of hi-rez files have slowed to a very deliberate pace. A near standstill, actually.

In any case, it should be reiterated that there are very, very few systems capable of reproducing 20 bits correctly. The best and most expensive resistors aren't currently built to those tolerances (at least, that's what my source was told directly by engineers at Vishay). So perhaps that shouldn't be the usual question, after all. 

[JerryM]

Okay, I get it - Sampling Theorem. Math is easy. 

But what do your ears tell you? Do you have a system capable of decoding a high resolution format? If so, do you have the ability to playback the high resolution source at the same time as the Redbook (or lessesr) source, so they can can be switched? Have you compared the two, yourself, on your own system, in your own listening room?

If not, mathmatical arguments mean very little. This is an audio website.

My ears tell me that high resolution digital audio sounds better than Redbook, period. Math really has nothing to do with that. 

Again, this is just my experience. If science has any place in it, so be it. Or not. 

Have fun,

Jerry

[Wayner]

Jerry,

I think the fundamental problem here is that the 44.1 format is being crucified here and it's a victim of circumstance. Again, for the most part (and there certainly are exceptions) the CD is for the masses, reduced in dynamic range to play even on a cheap boom box.

The hirez format has only one goal in mind, to satisfy the audiophile. So the real question in my mind is, are we all victims in a cruel plot, to believe that one medium is better then the other, knowing that one may have been under manipulation, for the purpose of satisfyingly the common masses.

That does not prove that one format is better then the other. It simply means that each has it's own purpose.

How can we scientifically evaluate each format, when the rules of engagement are different from the very beginning.

Wayner