[Darren G]

Another aside, but such a ton of valuable information here. Even if there is no agreement, this sub-forum, reading the insights from the audio engineers is cool crap how much depth and detail goes into mastering audio.

 

[71 dB]

Who the hell hears 4 dB sounds when exposed to peaks over 120 dB? Your ears will be still ringing the next day! That Conclusion 7 calculation is ridiculous.

 

[amirm]

And the noise floor of a typical living room is between 35 and 40dB, not 20 to 30.

That's just wrong. I have explained this many times to you. Yet you keep ignoring the science/psychoacoustics. Single number noise levels like this are useless. You must, must decompose them into their spectrum, convert them to equiv. sine wave levels, and then see how they compare. That is exactly what I did in my article on noise levels of our listening rooms. And that is based on peer-reviewed, journal of AES paper, written by ex-president of Audio Engineering Society. Again, he is Fielder in his paper:

And the figure:

As you see, even the average home has 10 db SPL of noise in the mid-frequencies where our hearing range is most sensitive (as shown by the solid line).

It doesn't matter if you agree or disagree on this topic as a whole. At least learn this bit of audio science and don't keep repeating objectivists myths of using one number to determine noise audibility.

Your SPL meter lies to you. It absolutely does. It doesn't have your brain or your two ears. In acoustic science, measurements can get you in trouble and fast. While said in a different context and purpose, Dr Toole says this best in his book:

This disagreement between what is measured and what is heard has been
the motivation for much scientific investigation of the acoustics of rooms, both
large and small. In some ways, our problems with rooms, especially small
rooms, began when we started to make measurements.​

I will say it again, please don't keep using single value SPL numbers as meaning anything with respect to audibility of noise. It is just junk forum objectivists banter that has no basis in real science.


Even worse in this context is the fact that we are in headphone forum. Headphones can block amazing amount of noise in those mid-frequencies. Once there, you can easily hear the faintest level of noise created by the equipment. Here is the data from my IEM, the etymotic E4SR:

That is a whopping 55 to 60 db of noise reduction at mid frequencies!!!

Furthermore, while stereo playback allows the sound in one channel to mask the noise in the other, no such thing happens with headphones as each ear hears what is intended for it.

So whatever argument you have here, simply does not hold when it comes to headphones.

Honestly, I have shown you all of this science and measurements before yet you repeat the same argument over and over again. Do you have any references to back your assertion and if not, why insist on it over and over again?

 

[Strangelove424]

How is the "loudest real life music" instrument measured? I'm assuming acoustic sources only, then which instrument and how was it mic'd up? Did they stick a mic directly into the drums, or rest it on the skin? These numbers are insanely high for a music listening scenario. 120db is way beyond the safe limit. To give you some perspective, here's a French air raid siren at 120db, or this car-mounted train horn. Who listens to music at 120db? They must be deaf. I want to meet that person. I'm sure they would ask me "what?" alot. BTW, 24 bit is artificially reduced to 123db SNR (20 bit equivalent). A true 24 bit dynamic range is an illegal signal and dangerous for equipment/hearing. So with an instant deafness-inducing 140db SNR ruled out, it's simply a choice between a painful 96db or tinnitus-inducing 120db.

 

[bigshot]

120dB is a VERY UNCOMFORTABLE LISTENING LEVEL. Find another cite that says you need 120dB of dynamic range to accurately reproduce recorded music. Happy hunting!

Honestly, if you knew what the numbers on the page represented, you would have a much better batting average at sorting out the wheat from the chaff.

 

[71 dB]

Honestly, I have shown you all of this science and measurements before yet you repeat the same argument over and over again.

Yes you have but you don't understand the relevance of the numbers you have shown. People don't hear the noise floor of 16 bit audio when they listen to music, not even when the track fades away. 16 bits would be enough even without dither and dither gives a few bits worth of extra practical dynamic range (up to ~120 dB!). In comparison the dynamic range of vinyl is 10 bits at best and vinyl lovers think even that is enough!

 

[amirm]

Yes you have but you don't understand the relevance of the numbers you have shown. People don't hear the noise floor of 16 bit audio when they listen to music, not even when the track fades away.

How did you determine that? Or is this based on some research I can read?

 

[JaeYoon]

I've said it before and I'll say it again. If you aren't listening to your music at 122dB, you don't need a dynamic range of 122dB. And the noise floor of a typical living room is between 35 and 40dB, not 20 to 30. Your numbers are wrong. If you get out a SPL meter and check for yourself what those numbers represent in real world sound, you'll see I'm right.

Can confirm, I installed a SPL meter app on android and can confirm 35 - 46 DB in my room.

 

[bigshot]

I don't think amirm owns an SPL meter. Even a cheap one would give you enough info to be able to sort out hooey like that. Have you tried getting a level on a loud listening level JaeYoon? It sure isn't 120dB!

 

[Don Hills]

When you look at an SPL meter in a "quiet" room, you're looking at the peak SPL at whatever frequency it happens to occur at. If you look at the graph posted by Amir, you can see that the SPL increases at lower frequencies, and that's what the meter is showing. It's also why you get a dramatic difference in the reading when you switch between A and C weighting. You should be using A weighting for hearing threshold type measurements because it more closely matches the ear's own sensitivity curve. It's still not close to a perfect match though. So in this case Amir is correct, you need to measure the SPL at multiple frequences and plot the curve to get a true picture of how audible the SPL in the room actually is.

 

[71 dB]

How did you determine that? Or is this based on some research I can read?

I have been listening to CDs for 3 decades and never have I heard the noise floor of 16 bit audio. All the noise I have heard is part of the recording and would be there no matter how many bits were used. I have also never heard anyone complain about hearing 16 bit noise floor. As an acoustic engineer I have some understanding of the practical demands of dynamic range in audio and anyone who has played with 16 bit audio in a wave editor, downsampled 24 bit to 16 bit using dither understands that 16 bit is enough.

The overall dynamic range of human hearing is 120-130 dB from hearing threshold to pain threshold, but not all of it is usable. If you are exposed to loud sounds, your hearing threshold raises temporalily. After having been some time in a silent environment you can hear silent sounds again. That's why you really need less dynamic range than 16 bit audio offers and why the "10 bits at best" dynamic range of vinyl seems to be enough for many.

I can understand why someone who knows nothing about digital audio or human hearing would think 24 bit consumer audio is better than (has benefits over) 16 bit consumer audio. I guess it's human nature to insist on "intuitive" concepts even when your undertanding and knowledge tells you otherwise. Companies making money on ignorance of people surely want to promote 24 bit high-res as "superior", but we who say 16 bit is enough have no such motives. Our only motive is telling the truth, because we feel it's our duty to use our knowledge that way against financial interests and lies.

 

[gregorio]

Noise in your recording comes from your speakers, i.e. point source. Noise that is in the environment in your room is diffused all around you. Putting aside the important bit that the spectrum of noise in our living room is anything but "white" (it is heavily biased towards low frequencies because walls and doors don't filter it as much), our hearing system is capable of distinguishing between point noise sources and diffused one.

Obviously my OP was simplified for brevity. It would have been entirely possible to write a short book just on noise, because it's a complex subject with a large number of variables: On the purely digital side of things we've got the variable of noise-shaped dither, which has been standard recommended practise for well over 15 years and first started being used over 20 years ago. On the recordings themselves we got the noise variables of mics, mic pre-amps and mic positioning, the noise of the recording location and the instruments and artists present in that location and then we have the noise induced by processing during mixing and mastering. Lastly, we've got the noise of the playback system, the DAC, amp, listening environment and transducers.

Nearly all of these variables are not just variable but highly variable and their interaction is complex. Yes, the typical environment noise of a listening environment is more pink-ish than white. On the other side of the coin, due to equal loudness contours at low levels, we're often not going to hear it as pink-ish, although that depends on where we live and the exact distribution of the noise. On the transducer side; yes, very well sealed IEMs might reduce listening environment noise by as much as 50dB or so, while open backed headphones might only reduce it by 10dB or so.

Despite my severe over-simplification, the basic point I was trying to convey still stands. Even if a certain combination of consumer equipment provides a listening noise floor of 0dBSPL, still 16 bit digital would allow for a peak level of 120dBSPL which, with such a low noise floor, would be way beyond comfortable and into the range of potentially damaging.

We measure the system level of noise, compare it to threshold of hearing, then determine the loudest real life music we can find, and the required dynamic range falls out of that.

No, that's what you're doing, it also appears to be what some scientists do BUT it is NOT what "we" do. Your apparent definition of "real life" music is bizarre, to say the least! Sure, if we place a mic a CM or so from many instruments, we could in theory record levels up 120dB or more. Heck, place a mic a CM or so from someone's mouth when they're speaking just a little loudly and we can get also easily get peak levels of 120dB+ ... BUT, put together your variables of: An extremely quiet listening environment, someone speaking a bit louder than normal and put your ear just a centimetre or two from their mouth and observe for yourself the results, or rather DON'T, DO NOT TRY THIS!! It's not a safe experiment, if it's a quiet environment NEVER allow someone to put their mouth very close to your ear and speak, unless they whisper very much more quietly than their normal speaking level! So, what is "real life" music? You typically place your ear a CM or a couple of inches from the instrument/s during a performance do you? How is that in any way "real life"??

In most scientific papers, the output levels and freqs are typically measured by a mic placed very close to the instrument/s and this is the max level/freq quoted. In real life, the situation is very different. In the case of most popular music genres, we do often record with a mic placed closely to the sound source BUT in order for it to sound subjectively pleasing, we then have to heavily process such recordings: EQ, artificial reflections, very significant level changes and various stages of compression to "shape" the transients. With acoustic music genres (classical, jazz, etc.), we tend to not apply heavy processing during mixing and mastering, instead, by positioning mics far more distantly we effectively achieve natural compression, transient shaping, input levels, EQ and reflections, due to absorption, and then all that's needed is just simply balancing the output of those mics during mixing/mastering with very little (or sometimes even no) additional EQ and compression.

In short, you're talking about the real life levels achievable with close mic'ing and adding this to the real life acoustic music recording technique of little significant processing BUT the problem is that this combination of real life situations do NOT actually occur in real life!

Additionally, you are ignoring some of the science here; while we have a large dynamic hearing range, it's not all usable at once. Just like with our sight, we can see in a near completely dark room and we can also see in bright sunlight but go from a very dark room into bright sunlight and what's the result?. Our hearing maybe capable of 120dB or more of dynamic range but the most optimistic scientific figures I've seen puts this at a max of about 60dB at any one time and I don't believe it's coincidental that commercial recordings almost never exceed this same, roughly 60dB, dynamic range. This brings us back to the content which is being listened to, it is not designed to test the limits of the human ear, it's designed as entertainment and to be comfortable.

That's just wrong. I have explained this many times to you. Yet you keep ignoring the science/psychoacoustics.

And I've explained to you in another thread all of what I've just said above and yet here you are again, stating and quoting exactly what you did to start with, completely oblivious to what you've been told! You quote only the science/psychoacoustics which supports your agenda (whatever that may be) and reference it to "real life", while deliberately ignoring other science, psychoacoustics and "real life" practicalities (which doesn't support your agenda) and then you accuse others of "keep ignoring the science/psychoacoustics"??! If you're going to quote "real life" then you HAVE to quote "real life", not just certain cherry picked aspects of real life which suit your agenda but which are NOT real life because you are omitting the other factors which would actually make it real life!

G

 

[71 dB]

60 dB test demonstrates a lot: Make 2 seconds of pink noise in a wave editor and attenuate the latter 1 seconds by 60 dB. Then play the noise in a loop: LOUD - quiet - LOUD - quiet - … and try to hear the quiet noise. Yes, you can make it audible by turning the volume up so that the LOUD part is uncomfortably loud. This is "vinyl" dynamic range. 16 bit digital audio even without dither does much much better and with dither perceptually even better than that! That's why 16 bits is enough, perhaps even 12 bit optimally used and dithered could be I think.

 

[OddE]

120dB is a VERY UNCOMFORTABLE LISTENING LEVEL. Find another cite that says you need 120dB of dynamic range to accurately reproduce recorded music. Happy hunting!

Honestly, if you knew what the numbers on the page represented, you would have a much better batting average at sorting out the wheat from the chaff.

-Just to put @bigshot 's post in further perspective:

I have just checked out a hydraulic power unit for my employer, ensuring all of the instrumentation we applied to it actually works. (It does! Coffee for everyone!)

This particular unit had three 270kW pumps running at full blast; that is sufficient energy to accelerate 150 metric tons of very expensive stuff away from the seabed, double haste. Standing between the pumps is a very physical experience - you can feel your whole body being pounded from the vibrations, and even wearing (proper!) hearing protection (foam plugs + peltors), the noise level is such that I find it slightly uncomfortable.

I took a reading. 119dB(A).

However, this is loudness wars taken to extremes; hardly much by way of dynamic range - so I could see myself listening to music with the occasional peak around the 120dB mark. If I had stuffed foam earplugs in my ears and then put Peltors over them, that is!

Now, all I need is to find someone who wants to record the sound of butterfly wings fluttering inbetween same pumps, and I've found a business case for 24 bit audio...

 

[TheSonicTruth]

Damn it! You can have FLAC the size of MP3! Freaking EPIC!!!
Oh wait. Sometimes you get artifacting ("herding_calls" sample) but not on brickwalled audio.
Yeah, I hear some noise on Daft Punk - Instant Crush. That "gangam style" is way too brickwalled and thus fraudy.
Upd. Using dithering can fix that noise though it occupies ~100 Kbps more. Still the noise can be an audiocassette fetish for some.

And this brings up a point: This whole thread ignores the elephant-in-the-room: The CONTENT. 16bit VS 24bit is a moot point if the master is hyper-compressed and has had the top 4-8 dB brick-wall-limited off of it!

Now I do realize that that has to be done mainly at the behest of the clients(the artist, the band, or the producer or label) that want volume level 10 loudness at a volume knob setting of 2, but it belays digital audio's ability to actually sound great.

Most of us on here have probably read that on-line article about the Nirvana 'Nevermind' 24bit high-rez downloads, and what was lost. But that example proves that there is no point in a high-res deliverable if the content itself is, under the marketing guise of 'remastering', treated to a steam roller and a lawn mower set too low!

That said, 24-32bit/96-192khz sampling will continue to be the obvious choice for tracking, mixing, and post, while 16/44.1 will still be more than adequate for deliverable(CD, download, etc) for the forseeable future.

Once both artists and engineers get past this addiction to everything in a song - from the lead vocals to the rhythm section to the melody and backing vocals - being dialed up to eleven, and digital's true dynamic range is seriously exploited, then we can start talking about 'higher res' deliverable formats.