Shouldn't we upgrade the 20-20 audible range ?!

[Sal1950]

You have been on this rodeo before, and whether lossy can be 'good enough' and why and how, has been hashed out extensively on QuadraphonicQuad, where they used to yuk about 'Dullby' (but gee, not so much, anymore).

The claim is not that lossy can never be distinguished from lossless. That's not a claim I would ever make. What matters are the conditions under which the difference can be heard, and the correct ways to determine that at it was heard. The average punter hasn't likely done the due diligence on those to so blithely dismiss lossy (including, specifically, Dolby Digital and MP3 codecs) as 'not good enough'.

So then I've yet to understand what your criticism of my simple two sentence post is?
Other than to be dis-agreeable , I'm done
.

 

[EJ3]

The vast majority of people have always listened on very poor quality equipment since recordings became available but that hasn't stopped there being many stellar quality recordings made since the 1950s at least. They tend to be popular with civilian enthusiasts too, assuming the music is accessible. How many copies has The Eagles Greatest Hits sold?- one of the best selling albums of all time, high DR.

There's also Aja, Brothers In Arms, Harvest...

The vast majority of people that I knew from 1971 (when I was 14) through the end of 2001 (when I left the USA for tropical islands until 2018, when I came back to my James Island, SC home [yes, the same one since 1964]), had very good to great audio gear. I am sad to say, that upon my return to the USA only 3 people that I know (1 is female) have very good audio gear. So, while great is available, most people that I know: are doing other things with their money.

 

[pablolie]

The vast majority of people that I knew from 1971 (when I was 14) through the end of 2001 (when I left the USA for tropical islands until 2018, when I came back to my James Island, SC home [yes, the same one since 1964]), had very good to great audio gear. I am sad to say, that upon my return to the USA only 3 people that I know (1 is female) have very good audio gear. So, while great is available, most people that I know: are doing other things with their money.

Great audio doesn't cost great money these days ... Just just priorities.

 

[Keith_W]

Maybe you could do a cost-benefit analysis to understand if we should choose a higher sampling rate or not.

Pros: we cater to the 0.000000001% of the world who can hear beyond 20kHz, who own speakers capable of reproducing sound at that range, and who listen to music with frequency content that goes that high, and who cares enough about high frequency reproduction to chase it as a worthwhile goal.

Cons: storage and transmission costs go up for EVERYBODY. This may not matter to you who has a paltry collection of FLAC's, but for music streaming services like Tidal, Amazon, Qobuz, etc. the cost will be enormous. They would literally have to double their storage to go from 48kHz to 96kHz. Then they pass the costs on to their consumers.

This will have knock-on effects for hifi manufacturers. Speaker manufacturers will start including super-tweeters, amplifier designers would need to make amps with wider bandwidth, DAC's and streamers will need to cater for higher sampling rates.

All that doesn't really bother me, I can simply choose to avoid purchasing such equipment if I wanted to. But higher streaming costs WILL affect all of us.

 

[Julf]

I don't know if it is true but the Sony guys had the demand that the disc should fit in the breast pocket of a shirt and at least must be able to play a full performance of a certain overture. That defined playtime and size and thus the amount of info that must be on that disc.

I can recommend Kees Immink: The Compact Disc Story

 

[solderdude]

Interesting read.

To recap the disc size was determined by Philips management and was based on the size of Compact Cassette size which was popular.
The final size ended up 0.5cm larger.
16 bit was decided because it was 2 bytes and was based on computer technology. Philips suggested 14 bits as it was not able to create a 16 bit DAC but only 14 bit. Using 4x oversampling 16 bit resolution could be achieved with as an added bonus a much simpler post filter (reconstruction filter was digital) where Sony produced a 16 bit DAC but had to use a very difficult to make analog reconstruction filter.
16 bit was chosen by Sony as it would be silly to 'throw away' 2 bits of 2 bytes.

The choice for 44.1kHz had to do with the existing PCM converters using computer tape and video recorders.
As 44.1 was easier to remember than 44.056 it was simply decided to go for 44.1kHz.

That would give enough resolution and audio bandwidth and bettered what could be achieved by any analog gear substantially (other than frequency range).
Given that vinyl and tape were already sufficient for great sound quality and more than 20kHz audio bandwidth is not audible anyway it seems a logical conclusion.

The CIRC method was invented by Sony and superior to what Philips came up with. It involved using 16kbyte RAM which was expensive in those days but the idea was that this would become cheaper (and so it did ... a LOT).

Using the 8 to 14bit encoding and the chosen disc parameters including control bits and other data (time, number etc) meant a 12cm disc could hold just under 80min which was more than any vinyl record could do. In practice this was limited to 74-75 minutes.

CD format was agreed upon and is what it is and won't be changed.
Now everyone that feels/thinks/suspects/believes/knows 44.1/16 is not enough can use any of the other digital formats they like and where there is a demand the market will follow. No need to whine about 16 bit not being enough or the 21kHz or so bandwidth not being enough as there are other formats that allow this.

The whole 20-20kHz bandwidth that is usually put in data sheets (without the all important cut-off points) is just a number game and there is no need to change that to ... say ... 10Hz to 30kHz or so or even 5Hz to 50kHz (without specifying cut-off points).

For archival purposes 96/24 is already more than enough and 192/24 is cheap today as well.

 

[EJ3]

Great audio doesn't cost great money these days ... Just just priorities.

That is true, but the fact remains that people go cheap (different than "inexpensive") and without the knowledge that they could have good to great for just a few $ more.
The negative perception created by the majority of youtubers that you "MUST" spend big money to get good to great audio, seems to be a big part of the issue.
It dissuades people from taking the time to check out the reality (which, barring them having friends that tell them about this site), they do not take the time to hunt it down. They just presume that the "snake oil" of you must spend big to have good to great audio is true.
Because there is so few promoting the truth that it can be relatively inexpensive.
But, there is also the consumerism pressure to buy the cheapest junk that you can (I got this soo "CHEAPLY" [no matter that it is, in fact, landfill junk in 6 months]) but you got a deal on it: mentality that prevents many from even checking for better quality of sound.
So: we need to try to spread the word that you can, with a bit of due diligence, get great sound inexpensively.
Partially by pointing them to this site (perhaps even by showing it to them and a bit about navigating it), as well as perhaps some other relevant sites.
I have managed to get one of my friends doing so (he has always had an interest in good to great audio but did not know where to look after the disappearance of great audio stores from our city). I had been overseas from 2001-2018 & one day suggested we go to one of our former audio stores that we had bought things from, he said "There are no more audio stores to go to anymore, only big box stores like .... ...". So, I turned him on to here. And he has gotten a few things because he now knows a place or 2 to check before he buys.
And he has passed that experience on to at least one other person.
But, many: are not receptive to the idea of being able to have great audio inexpensively (or other things that are great, inexpensively: [even when you show them it is so]).
Such is the consumerism mentality ingrained, that it is hard for many to think of research for value of buying "quality inexpensively" as opposed to kneejerk buying trendiness.

 

[welwynnick]

A few questions for you. Suppose you could set a new standard....
1: What research shows 44.1/16 LPCM is not 'transparent' ?
2: What is meant with 'transparent' (as in what is the or your definition) ?
3: If science would find 44.1/16 insufficient these days what would be sufficient... 60/20 or 88.2/16 or 44.1/24 or 48/24 or 96/24, DXD ?
4: If there were a need for physical storage of an album what format/type would that be ?
Consider that around 50 years ago when development started the best that could be produced was 16 bit converters (and even they did not reach true 16 bit performance).
Likewise the optical technology that could be made at that time had practical limitations.
CD thus was a compromise based on what was known about minimum requirements for audio and what was feasible to produce at that time using digital technology for home audio.
Consider that there was a physical size and playtime constraint as well as optical method limiting the CD format.
I don't know if it is true but the Sony guys had the demand that the disc should fit in the breast pocket of a shirt and at least must be able to play a full performance of a certain overture. Turns out it was a rumor. That defined playtime and size and thus the amount of info that must be on that disc.
At that time there were no micro-SD cards nor streaming that could hold tens/hundreds of albums in a much higher resolution nor did anyone have an idea that 30-40 years later people would play music on a personal device with unlimited access to information and nearly all music one could ever listen to in a lifetime.
So... what would you set as specs if it were up to you and what would be the rationale for choosing that ?

I believe that Sony & Philips launched compact disc when reliable and reproduceable laser disc technology had advanced to the point where the full 74 minutes of Beethoven's 9th symphony could be recorded, using a format that was considered transparent at the time, onto a disc that could fit into a player the size of a car radio.

Of course, storage and distribution technology continued to improve, and that was predictable. I dearly wish that Sony & Philips had anticipated how enduring their standard might be, and hadn't been so impatient. I've worked in complex electronic systems for a long time, and one of the lessons is that if you want a standard to be robust and effective , you need to design it to do more than the minimum requirements, and consider reliability, growth and evolution. CD met it's requirements by the skin of it's teeth. Allowing just 2kHz for the reconstruction filter transition was asking too much - they should have allowed a much bigger margin. I would have gone for 88.2 or 96kHz like DVDA, but even adopting 48kHz at the time might have been enough to quash the subsequent "CD isn't transparent" bickering. Nor was the medium robust enough to avoid bit errors being an issue.

It's pretty clear that 16/44 was transparent in 1982, but that was more down to the quality of the supporting electronics, rather than the recording format itself. How else could you prove that it wasn't transparent if the format wasn't the bottleneck? In this context, transparent meaning that it's not possible to distinguish one system from another under controlled level matched listening with any material and any replay system with high, say 95%, confidence. Scientifically, if you want to do a critical assessment of something, your measurement should be better than your target (rather than worse, in this case).

Over the years I've been collecting examples of controlled listening test reports that do show 16/44 isn't transparent. Goldensound was only the first one that I searched for in this thread. I thought it was interesting because while it didn't necessarily prove what it set out to do, which was to demonstrate audible differences between DACs, I think it did show that 20kHz isn't enough for transparency. I've got a lot of other test reports that show the same thing. In particular they show that while ultrasonics are inaudible in isolation, they are detectable when played on top of the corresponding audible sound. It's counter-intuitive.

I'll read the thread and see if they've already been discussed. Here's a hint though - some of them are on HydrogenAudio, where ABX DBT is king.

 

[solderdude]

I believe that Sony & Philips launched compact disc when reliable and reproduceable laser disc technology had advanced to the point where the full 74 minutes of Beethoven's 9th symphony could be recorded, using a format that was considered transparent at the time, onto a disc that could fit into a player the size of a car radio.

See post #1026.. believing is not knowing (I heard incorrect rumors as well and perpetuated them)

Of course, storage and distribution technology continued to improve, and that was predictable. I dearly wish that Sony & Philips had anticipated how enduring their standard might be, and hadn't been so impatient.

How many years should they have waited (after the approx 10 years they were into and needed to bring something new) ?
5 years, 10 years ?
What would have changed ?

I've worked in complex electronic systems for a long time, and one of the lessons is that if you want a standard to be robust and effective, you need to design it to do more than the minimum requirements, and consider reliability, growth and evolution.

Have been in electronics development as well and often built-in features that might be useful in the future but were never used. Instead something entirely different was marketed later.

CD met it's requirements by the skin of it's teeth. Allowing just 2kHz for the reconstruction filter transition was asking too much - they should have allowed a much bigger margin.

That's your assessment and not the general consensus. A good and sharp enough digital filter at 44.1kHz is more than sufficient. Where is the proof that it isn't.

I would have gone for 88.2 or 96kHz like DVDA, but even adopting 48kHz at the time might have been enough to quash the subsequent "CD isn't transparent" bickering. Nor was the medium robust enough to avoid bit errors being an issue.

This was not an option at that time.
Why would a higher bit-rate or -depth be in the RB at that time which would severely impact playing time ?

It's pretty clear that 16/44 was transparent in 1982

If it was transparent 'then' why would it not still be now ?

Scientifically, if you want to do a critical assessment of something, your measurement should be better than your target (rather than worse, in this case).

Audibly transparent is the same now as it was in the 80's. Back then bandwidths of 40kHz were already possible and existed. The dynamic range back then also was already enough with good tape and companders.
Nah... transparent did not become more transparent and humans have not evolved to hear higher frequencies (at a young age) over the last few decades either.

Over the years I've been collecting examples of controlled listening test reports that do show 16/44 isn't transparent. Goldensound was only the first one that I searched for in this thread. I thought it was interesting because while it didn't necessarily prove what it set out to do, which was to demonstrate audible differences between DACs, I think it did show that 20kHz isn't enough for transparency. I've got a lot of other test reports that show the same thing. In particular they show that while ultrasonics are inaudible in isolation, they are detectable when played on top of the corresponding audible sound. It's counter-intuitive.

I'll read the thread and see if they've already been discussed. Here's a hint though - some of them are on HydrogenAudio, where ABX DBT is king.

Please don't bring up Cameron as it is unknown what he heard and why and there was no confirmation under better controlled conditions.
Also he did not claim audible better sound quality and there was no '20kHz vs 30kHz' but at best 21kHz vs 22kHz.
It did NOT prove anything about HiRes (>48/24) sounding 'more transparent' than transparent.

Similar to your square-wave experiment. You have heard a difference but did not measure the actual signal but just argued it and even incorrectly.
YOU need to prove what you heard and show the measurements of the signals and AB test results for instance.

 

[RandomEar]

Goldensound was only the first one that I searched for in this thread. I thought it was interesting because while it didn't necessarily prove what it set out to do, which was to demonstrate audible differences between DACs, I think it did show that 20kHz isn't enough for transparency.

In the thread about this on ASR, a number of relevant questions have been brought up. Sadly, Goldensound himself doesn't post here anymore and his friend, who was present during the test, could not clear them all up. Necessary clarifications are still missing and follow-up tests to determine the actual cause of how he may have heard or not heard sound above 20 kHz have been promised but never delivered. See this post for the last update. So as it stands right now, I see this as a bad example because important details about how the test was performed are missing. Maybe it was just IMD in the audible range.

However, there's good research about the upper limits of hearing like Takeda 1992, which show that for young children, it's above 20 kHz. At age 15, that is still the case for about 10% of the population. At 26 (age of Cameron when the test was conducted), that number will be very close to zero according to the data from Takeda - but it might not be absolute zero. It would be fine by me if we extend the common range by ditching round numbers and end up with "18 to 21.4 kHz" or something like that. For the vast majority of the population and certainly everybody above 30, it wouldn't matter. But for some younger listeners, it might.

In general, I would feel comfortable with having a bit of additional headroom in the recording format. So 48 kHz and maybe 20 bits sounds good to me (Notice the pun? Sounds good. Ha!). That would be covered by DVD audio (48 kHz, 24 bit), which never really took off. CD audio simply proved to be good enough for 98% of situations and listeners.

I've got a lot of other test reports that show the same thing. In particular they show that while ultrasonics are inaudible in isolation, they are detectable when played on top of the corresponding audible sound. It's counter-intuitive.

I'll read the thread and see if they've already been discussed. Here's a hint though - some of them are on HydrogenAudio, where ABX DBT is king.

I remember the Reiss meta-analysis from 2016 being discussed on ASR. I think this post sums up most of the problems with it pretty well. I have not seen other convincing evidence around here, but maybe I missed it.

 

[EJ3]

In the thread about this on ASR, a number of relevant questions have been brought up. Sadly, Goldensound himself doesn't post here anymore and his friend, who was present during the test, could not clear them all up. Necessary clarifications are still missing and follow-up tests to determine the actual cause of how he may have heard or not heard sound above 20 kHz have been promised but never delivered. See this post for the last update. So as it stands right now, I see this as a bad example because important details about how the test was performed are missing. Maybe it was just IMD in the audible range.

However, there's good research about the upper limits of hearing like Takeda 1992, which show that for young children, it's above 20 kHz. At age 15, that is still the case for about 10% of the population. At 26 (age of Cameron when the test was conducted), that number will be very close to zero according to the data from Takeda - but it might not be absolute zero. It would be fine by me if we extend the common range by ditching round numbers and end up with "18 to 21.4 kHz" or something like that. For the vast majority of the population and certainly everybody above 30, it wouldn't matter. But for some younger listeners, it might.

In general, I would feel comfortable with having a bit of additional headroom in the recording format. So 48 kHz and maybe 20 bits sounds good to me (Notice the pun? Sounds good. Ha!). That would be covered by DVD audio (48 kHz, 24 bit), which never really took off. CD audio simply proved to be good enough for 98% of situations and listeners.


I remember the Reiss meta-analysis from 2016 being discussed on ASR. I think this post sums up most of the problems with it pretty well. I have not seen other convincing evidence around here, but maybe I missed it.

20 Hz is actually almost exactly between E and E flat, a couple of cents closer to E flat.
Bösendorfer makes pianos that go down to C0, a frequency of (nominally) 16.3516 Hz. 16 Hz is the speed of 32nd notes at 120 quarter notes per minute.
Note that a piano sound is not a sine wave, so even if the fundamental frequency were below the human hearing limit, we still hear the overtones and other higher frequency components contained in the sound.
The Boardwalk Hall Auditorium Organ in Atlantic City and the Sydney Town Hall Grand Organ are two of the very few organs that have a 64 foot pipe. These produce subsonic fundamentals of about 8 Hz (C−1) (along with harmonics). Rather than what we usually think of as "sound," they are perceived as physical sensations like pressure and rumbling.
C0 (about 16 Hz) is featured in the finale of Saint-Saëns' Symphony No. 3, also known as the "Organ Symphony" (1886).
So: maybe 8Hz or 16Hz should be the lower limit?

 

[RandomEar]

Well, if you can't hear it but just feel the pressure, it's getting difficult to judge. You would still need a sub with insane low level extension and power to reproduce that feeling. Or maybe you could add a shake table to your room as a suitable transducer...

 

[Mnyb]

Well, if you can't hear it but just feel the pressure, it's getting difficult to judge. You would still need a sub with insane low level extension and power to reproduce that feeling. Or maybe you could add a shake table to your room as a suitable transducer...

Furniture intermodulation anyone ?

 

[Mnyb]

If return to the actual origianl question "Shouldn't we upgrade the 20-20 audible range ?!" i say 0-20 ,given that infrasound can be felt with the body .

so if 20 kHz can be sufficiently encoded we are done in that regard and its seems to me that this is the case ? so that is transparent to human listeners but electronic equipment can do better some speakers works up to 30kHz ( for the pleasure of your pets ) amps even more in some cases .
There are good enough filter implementations for both 44.1 and 48 ( recording at those i rate i don't know maybe best avoided today ? and OT )

Bit depth is not in the original topic but no practical electronics actually gives 144dB as in 24 bit so there are no "real" 24 bit equipment.
And I guess that the bottom 8 bits are equivalent to random noise in most recordings ? with some exceptions for very modern classical recordings or pure electronic music .

Q transparent to who ? 24/96 is better than any human but also better than most speakers and the bit depth is better than all electronics ?

Studio equipment and DAW have adopted floating point internally to avoid that any data representation is a limitation and are transparent to every standard imaginable ?

Down sampling is for human consumption of the end result .

 

[danadam]

Bösendorfer makes pianos that go down to C0, a frequency of (nominally) 16.3516 Hz. 16 Hz is the speed of 32nd notes at 120 quarter notes per minute.
Note that a piano sound is not a sine wave, so even if the fundamental frequency were below the human hearing limit, we still hear the overtones and other higher frequency components contained in the sound.

I'm not sure the fundamental is actually present:

The Truth about many "Audiophile" Piano Recordings

Reproducing the sound of a piano indistinguishable from the real thing is a hard ask, so I wouldn't be expecting it. Some people have taken up the challenge. In France, Georges Cabasse and his closer associates, sound engineers Bernard Neveu and latter on Philippe Muller, have repeatedly made...

www.audiosciencereview.com

C0 (about 16 Hz) is featured in the finale of Saint-Saëns' Symphony No. 3, also known as the "Organ Symphony" (1886).
So: maybe 8Hz or 16Hz should be the lower limit?

And what does it mean exactly? It is already possible to go that low, at least as far as the medium (CD) is concerned:

BASS!

hope you didn't measure the YT version,bass there is non existent I doubt that it's much different from the (SA)CD. That's how it looks like in REW's wav's analyzer: What length of the intro is analyzed there? And why on SPL scale? It's pretty useless in this case. If you switch to dBFS...

www.audiosciencereview.com

or the symphony you mentioned: www.youtube.com/watch?v=fZZmvBR077Y&t=2170s

 

[welwynnick]

Please don't bring up Cameron as it is unknown what he heard and why and there was no confirmation under better controlled conditions.

Tit-for-tat posting never goes anywhere good, but I'd like to defend Cameron. He was as transparent as he could be in his testing, and he did post the test material, and even a video of himself doing the test. Admittedly I don't think he proved DACs sound different unless you use different filters, but I thought he got an interesting and useful result if you just treat it as "how high can you hear?" and he passed that with a high degree of confidence.

Bit depth is not in the original topic but no practical electronics actually gives 144dB as in 24 bit so there are no "real" 24 bit equipment.

To be fair to Topping, ASR measured 143dB DR for the B100, and L7 measured 144 for the L70.
There aren't any DACs that good, though.

 

[RexrothPigeon]

Absolutely, do folks think any of these artists released these recordings not knowing they were squashed to 0dr?
There's 4 or 5 more pages of this garbage. LOL
View attachment 542084

hahaha that's amazing. fwiw at least some those artists are harsh noise musicians, whose defining technique is slamming max sonic density at all times, dynamics be damned.

 

[EJ3]

Furniture intermodulation anyone ?

Well, I guess that would be up to the person listening/perceiving to get the gear needed to do so, if they wanted, Just as people who want to level match 21.5 Hz will likely need to buy a supertweeter.
My amps can certainly reproduce it (and I could retune my custom sub boxes to probably do 16Hz [8 is unlikely, though]).

I'm not sure the fundamental is actually present:

The Truth about many "Audiophile" Piano Recordings

Reproducing the sound of a piano indistinguishable from the real thing is a hard ask, so I wouldn't be expecting it. Some people have taken up the challenge. In France, Georges Cabasse and his closer associates, sound engineers Bernard Neveu and latter on Philippe Muller, have repeatedly made...

www.audiosciencereview.com

And what does it mean exactly? It is already possible to go that low, at least as far as the medium (CD) is concerned:

BASS!

hope you didn't measure the YT version,bass there is non existent I doubt that it's much different from the (SA)CD. That's how it looks like in REW's wav's analyzer: What length of the intro is analyzed there? And why on SPL scale? It's pretty useless in this case. If you switch to dBFS...

www.audiosciencereview.com

or the symphony you mentioned: www.youtube.com/watch?v=fZZmvBR077Y&t=2170s
View attachment 542208

I wasn't talking about the recording of the piano. I said that is the note that the piano produces.
And the symphony (having never heard it myself), I'm going by the musical notations.
Recording it and then reproducing that recording are a different problem.
From my amps perspective and my CD players perspective, that low note is not an issue.
But it is also not covered by the 20Hz-20KHz idea of FR.

 

[Julf]

I believe that Sony & Philips launched compact disc when reliable and reproduceable laser disc technology had advanced to the point where the full 74 minutes of Beethoven's 9th symphony could be recorded, using a format that was considered transparent at the time, onto a disc that could fit into a player the size of a car radio.

It was considered transparent then, and it is considered transparent now. Human hearing hasn't improved.

Of course, storage and distribution technology continued to improve, and that was predictable. I dearly wish that Sony & Philips had anticipated how enduring their standard might be, and hadn't been so impatient. I've worked in complex electronic systems for a long time, and one of the lessons is that if you want a standard to be robust and effective , you need to design it to do more than the minimum requirements, and consider reliability, growth and evolution.

They did just that. The state of art the time was the BBC FM transmissions - that were transmitted on a digital backbone network using 14 bit sampling at 32 kHz, compressed into lossy 10-bit NICAM.

CD met it's requirements by the skin of it's teeth.

Wrong.

Nor was the medium robust enough to avoid bit errors being an issue.

No medium is 100% robust. CD is amazingly robust.

It's pretty clear that 16/44 was transparent in 1982, but that was more down to the quality of the supporting electronics, rather than the recording format itself.

The supporting electronics were more than good enough even back in 1982. Much better than even the best modern recordings.

Over the years I've been collecting examples of controlled listening test reports that do show 16/44 isn't transparent.

Yes, we see that it is an obsession for you, but in science it is called "cherry picking" and is considered a cardinal sin in academic research.

 

[welwynnick]

but in science it is called "cherry picking" and is considered a cardinal sin in academic research.

Absolutely, hence the hypocrisy of this thread. Everybody rejects the evidence that doesn't support their own beliefs.