Author Topic: Can you truly make 16bit audio sound better by upping it to 32bit?  (Read 16134 times)

0 Members and 1 Guest are viewing this topic.

Offline retrolefty

  • Super Contributor
  • ***
  • Posts: 1648
  • Country: us
  • measurement changes behavior
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #50 on: August 18, 2017, 10:19:35 pm »
 It's a hopeless debate that has gone on for many (to many) decades. While some might rightfully demand proof that someone actually hears what they think and say they hear. It's so subjective to turn off most technically knowledgeable people, but too emotionally strong in 'golden ear' believers to not say "you can't measure what I hear and feel".

 I firmly believe that writing, producing, and mastering of music content is an ART and not limited to or by engineering principles.

 On the other hand I feel that music reproduction in one's home falls purely the domain of good engineering principles. Speakers and room layout will always be the limiting technical factors and dwarf any of the electronics including magic audio capacitors,$5K triode class A vacuum tube amps, $1K power and speaker cables, etc.

 There is no way out or hope for this topic................

   
« Last Edit: August 18, 2017, 10:22:16 pm by retrolefty »
 

Offline WastelandTek

  • Frequent Contributor
  • **
  • Posts: 609
  • Country: 00
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #51 on: August 18, 2017, 11:15:58 pm »
Well, I think it is very understandable for people to be uncomfortable with the idea that they are living in a simulation.
I'm new here, but I tend to be pretty gregarious, so if I'm out of my lane please call me out.
 

Offline DrGeoff

  • Frequent Contributor
  • **
  • Posts: 793
  • Country: au
    • AXT Systems
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #52 on: August 18, 2017, 11:33:49 pm »
Well, I think it is very understandable for people to be uncomfortable with the idea that they are living in a simulation.

And the lack of comfort would be part of the simulation since people could not have ideas beyond those specified by the simulation parameters.
Was it really supposed to do that?
 

Offline CatalinaWOW

  • Super Contributor
  • ***
  • Posts: 5160
  • Country: us
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #53 on: August 19, 2017, 03:41:19 am »
The fact that we don't have the hard science behind all of audio doesn't mean those things don't exist.  It just means we are more like Faraday, performing experiments and drawing conclusions before Maxwell pulled it all together.

I can see this even though I was no golden ear in my youth, and far less so now.  I never could keep records clean enough to avoid pops, snaps and other objectionable behavior.  So I transcribed all of my library to .mp3 files, and transcribe new old stock records as I find and buy them.  While I do use a high sample rate and bit depth, it is not because I can really hear a difference.

Where I see funny phenomena is as I go through these recordings with Audacity to remove the noise artifacts.  The automatic noise removal functions are not very effective, either leaving the noise in place, or distorting the sound depending on sensitivity settings so this cleanup is tedious and manual.  I listen.  When I hear an artifact I stop playback and go back to find it and edit it out.  As I am scrolling through the sound trace looking for the guilty party I often see clear discontinuities and other disturbances that turn out not to have generated the offending noise or any noise that is audible to me.  The actual guilty party is often just a little wiggle on the sound.  The total energy of the difference from the desired sound signal is far smaller than many of the other much larger artifacts.  Artifacts which cause similar auditory discomfort to me have physical amplitudes which vary by as much as 30 dB.  This inability to select visual artifacts which will cause objectionable noise makes the tedious process even more difficult.  An album with only a couple of pops and static may take three times the playing time to clean up.  One with higher noise levels can take days.  Such effort is justified only on a very enjoyable and rare recording.

In attempts to automate this process I have tried to identify characteristics of "objectionable" or "audible" noise with little success.  It is not a simple function of dV/dt, or frequency content or other classic signal processing measures.  Higher listening levels make the noise more detectable and repetition also makes them more detectable, but I have been unable to quantify these observations.  Maybe if I read the thousands of papers on psychoacoustics I would make more progress.  In the meantime I have no problem agreeing that there are unexplored wilds in the world of audio.
 

Offline b_force

  • Super Contributor
  • ***
  • Posts: 1381
  • Country: 00
    • One World Concepts
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #54 on: August 19, 2017, 08:04:12 am »
The fact that we don't have the hard science behind all of audio doesn't mean those things don't exist.
Actually, the fact is, that there is enough hard and good science about it.

But for some reason people don't want to dive into it.
Almost like they refuse to face the truth.
One of the reasons why it's complicated is because there is a lot of psychology involved. A type of science which is not always being considered as real science by engineers and tech scientists.

Offline Alex Nikitin

  • Super Contributor
  • ***
  • Posts: 1148
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #55 on: August 19, 2017, 09:52:33 am »
Actually, the fact is, that there is enough hard and good science about it.

Actually, the fact is, there is not much "hard and good" science about it. What pretend to be a "science" is not based on measurements and numbers, and what in that area of knowledge is pretend to be based on measurements and numbers, is not a science but mostly a con art.

But for some reason people don't want to dive into it.
Almost like they refuse to face the truth.
One of the reasons why it's complicated is because there is a lot of psychology involved. A type of science which is not always being considered as real science by engineers and tech scientists.

Psychology is not a science, it is a lot of crap mostly made up on a spot . You can not measure something if your tools are not suitable for that measurement, plain and simple. You can not measure 10mV DC voltage with a needle meter for 1kV full scale, and that is what most of this "science" about human audio perception tries to do. No real positive data except on the level "OK, there is a sound and now there isn't"  :-DD . And without a positive baseline established all the negative data ("this difference is impossible to hear according to our very scientific experiments") worth nothing.

Cheers

Alex
 

Offline A Hellene

  • Frequent Contributor
  • **
  • Posts: 602
  • Country: gr
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #56 on: August 19, 2017, 12:51:08 pm »
Psychology... Ah, yes... That NON-scientific masterpiece of our times, based on anything but rock-solid scientific facts...

Meaning no disrespect at all to our counterparts of this specific department, this will be a version of a rant I have had a decade ago in another discussions board about my personal opinion on Psychology.

But, before that, please watch the documentary called 'Psychiatry: An industry of death':
Psychiatry: An industry of death (2006).
Beware, though, because this documentary has its fare share of bits and pieces of the usual propaganda; as it always happens with the vast majority of any release for public consumption...

Well, let's take apart the fancy term, 'chemical imbalance.' Should not it be mandatory for anyone that 'diagnosed' us with some kind of 'chemical imbalance' to inform us what a 'chemical balance' really is? With documented factual numbers, if possible? No one can do that! Because this is yet another buzz-word for the well-meaning but gullible also masses. Just consider that this specific 'scientific department' considers even the healthy laughter as a disease that needs to be addressed with psychotropic poisons that will most probably cause a permanent brain damage to their recipients... You can still find in Youtube televised advertisements of the fifties about Lobotomy (yes, that good old psychiatric practice with an ice-pick...), in the context of: Is your wife getting tiresome when you return home from work and she does not let you watch your favourite TV programme? Make her a gift you will both appreciate: A Lobotomy! Yes, this actually was a TV commercial message of that era, before that medical practice of lobotomy that left Rosemary Kennedy permanently disabled for life; thus the invention of the Special Olympics in 1962 in her memory...

So, we have our therapists (sic) to confess our personal dramas (and facts) to, since people are not so pious today to be confessing to the priesthood anymore, while a formal inquisition is not such an appealing procedure for either party to brag about. As for the therapists themselves, they will be happy to be filled in with the latest gossip and to be payed for that, along with the prescription of useless and dangerous drugs that will allow them to earn their percentage from the Big Pharma. In my humble experience as an observer, people taking psychotropics will never overcome their problems because their self-healing ability will actually be reduced by the medication: only the pharmaceutical industry, the "therapists" and whoever gathers personal information or spies on the victims will benefit from this charade...

What about the so-called 'mental disorders'? Are they for real? There are a few hundred distinctive names for mental disorders that have been made up; but if you would go to several different 'professionals' and tell them the same exactly story, you would probably end up getting several different kinds of 'chemical- or 'electrical-imbalance' diagnoses and several different drugs for their 'cure': This is NOT science; there is another pejorative term for those practices. Until recently, there still was the Four Humors myth alive, since the days of Hippocrates, 2500 years ago... As for the inconsistency of the 'diagnoses,' if you are on insurance and they pay, you are going to be labeled indefinitely. Insurance has certainly perverted the system by forcing the therapist to make a diagnosis in order to get payment.

Trying to take a look in depth, why are the religious people (who believe in things like angels and demons) considered to be healthy, but if someone dares to believe in elves and goblins is considered to be mentally disordered and schizophrenic? Neither of them has, or can ever positively prove the existence of the angel/deamon or the elf/goblin entities they believe in; both of them are equally invalid. Not to mention that in some special cases, where people in high places have to explain their beliefs in elves and fairies are no more considered to be mentally disordered: they are just been called Pagans, and that is alright... Point taken, or should I continue?

As a result, I would like to point out that most people are too scared to look at the whole truth because in their vast majority they think with their feelings instead of their reason. No need to be rational if it already feels good... This is the exact mental (dys)function that is being taken advantage of by every kind of human manipulators; everywhere; ever...


-George
« Last Edit: August 19, 2017, 01:08:08 pm by A Hellene »
Hi! This is George; and I am three and a half years old!
(This was one of my latest realisations, now in my early fifties!...)
 

Offline b_force

  • Super Contributor
  • ***
  • Posts: 1381
  • Country: 00
    • One World Concepts
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #57 on: August 20, 2017, 04:59:45 pm »
Actually, the fact is, that there is enough hard and good science about it.

Actually, the fact is, there is not much "hard and good" science about it. What pretend to be a "science" is not based on measurements and numbers, and what in that area of knowledge is pretend to be based on measurements and numbers, is not a science but mostly a con art.
Than you have A LOT of reading to do, good luck!
I should start with an AES membership first.

Quote
But for some reason people don't want to dive into it.
Almost like they refuse to face the truth.
One of the reasons why it's complicated is because there is a lot of psychology involved. A type of science which is not always being considered as real science by engineers and tech scientists.

Psychology is not a science, it is a lot of crap mostly made up on a spot . You can not measure something if your tools are not suitable for that measurement, plain and simple. You can not measure 10mV DC voltage with a needle meter for 1kV full scale, and that is what most of this "science" about human audio perception tries to do. No real positive data except on the level "OK, there is a sound and now there isn't"  :-DD . And without a positive baseline established all the negative data ("this difference is impossible to hear according to our very scientific experiments") worth nothing.
That's the most simplistic definition of 'science' I have ever seen.

Psychology is based on possibilities, very much like physics is.
Nothing more than describing certain behaviors. You don't necessarily need any measuring tools for that even.
Countless examples in physics that work exactly the same way.
Unless you still believe in black and white or discrete results, but than you have been sleeping in a cave for more than 80-100 years.

Psychology also goes MUCH further than these very simplistic given examples.

Offline Alex Nikitin

  • Super Contributor
  • ***
  • Posts: 1148
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #58 on: August 20, 2017, 06:11:59 pm »
Than you have A LOT of reading to do, good luck!
I should start with an AES membership first.

I've considered an AES membership years ago. Went to several AES meetings in London, read the materials, etc . Lost interest after a while, as found it just a waste of time.

That's the most simplistic definition of 'science' I have ever seen.

Psychology is based on possibilities, very much like physics is.
Nothing more than describing certain behaviors. You don't necessarily need any measuring tools for that even.
Countless examples in physics that work exactly the same way.

Please name one example from physics.

Cheers

Alex
 

Offline b_force

  • Super Contributor
  • ***
  • Posts: 1381
  • Country: 00
    • One World Concepts
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #59 on: August 20, 2017, 07:29:19 pm »
ALL physics are based on statistics.

Most obvious are quantum physics, but even the famous F=m*a is based on statistics.
In other words, there is a possibility that something will 'fall up' instead of down for example.
The change is only so extremely small that it's extremely unlikely you will see it.

Offline Alex Nikitin

  • Super Contributor
  • ***
  • Posts: 1148
  • Country: gb
  • Femtoampnut and Tapehead.
    • A.N.T. Audio
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #60 on: August 20, 2017, 07:55:04 pm »
ALL physics are based on statistics.

Most obvious are quantum physics, but even the famous F=m*a is based on statistics.
In other words, there is a possibility that something will 'fall up' instead of down for example.
The change is only so extremely small that it's extremely unlikely you will see it.

Oh, so you've meant "probabilities", not "possibilities". Physics deals with probabilities by repeating the experiment multiple times and reaching a very predictable result. Essentially, you've just proved that psychology of perception is not a science, as this approach is not possible there. Statistics is just a tool and in audio perception experiments it is used mostly to cover the fact that there aren't any decent facts  ;) .

Cheers

Alex
 

Offline IanMacdonald

  • Frequent Contributor
  • **
  • Posts: 943
  • Country: gb
    • IWR Consultancy
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #61 on: August 22, 2017, 06:21:49 am »
@A Hellene - thought I should set you straight that psychology and psychiatry are two entirely different disciplines.

Psychology is about trying to understand the 'software' of the human mind.  Psychiatry is about treating mental health problems with drugs, on the principle that all such issues are a product of chemical imbalances in the  brain.

Psychologists freely admit that they don't understand the human mind all that well, but at least their approach isn't like hitting the computer in the hope of curing the fault.
 
The following users thanked this post: frozenfrogz

Offline A Hellene

  • Frequent Contributor
  • **
  • Posts: 602
  • Country: gr
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #62 on: August 22, 2017, 08:54:44 am »
Of course, Ian! What you very well described are the claims of those departments.

What I stated, in my message above, is what their actually practices are; and how scientific these practices really are (or have become), especially after the involvement of the banking sector (see: insurance companies) in their claimed 'science'...

Again, I am not criticising any well-meaning individuals involved in these questionable practices, since this is what they have officially been taught in order to make a living...

I am merely expressing my distrust in this specific pseudoscientific department that has been established by the 'Big Pharma' (a huge money-making and lobbying industry), since when money talks morals (as well as true science, in this case) disappear.


-George
Hi! This is George; and I am three and a half years old!
(This was one of my latest realisations, now in my early fifties!...)
 
The following users thanked this post: IanMacdonald

Offline IanMacdonald

  • Frequent Contributor
  • **
  • Posts: 943
  • Country: gb
    • IWR Consultancy
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #63 on: August 22, 2017, 03:30:47 pm »
A useful comparison could possibly be made between psychology today and alchemy a few hundred years back.  Now, a large part of alchemy was pure BS, but then the alchemists did lay down the principles of good experimental practice that are still used in every lab today.  If there had been no alchemy, then likely there would have been no chemistry.

So, I don't think it's fair to knock people who are trying to figure something out which we have little understanding of. People in that position are bound to make some wrong conclusions before they get it right. For example, Freud's notion that all human behaviour is sexually motivated is largely discredited today. It was a reasonable hypothesis at the time though. 
 

Offline CatalinaWOW

  • Super Contributor
  • ***
  • Posts: 5160
  • Country: us
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #64 on: August 22, 2017, 05:34:33 pm »
To further your example.  Although Freud's belief that sexual drives were at the root of everything, I suspect most today would support a modified version - that sexual drives influence much of our behavior.  That would be typical of most science trajectories.  A refinement of prior theories as better measurables come along and show the holes in earlier theories and allow better ones to develop.

While much audiophoolery is hogwash, the example I provided earlier is a case where I can't explain an audio result based on the best tools I have from science - as an electrical engineer with advanced degrees and years of experience using and applying frequency response, power, dV/dt and other elements which supposedly explain it all.  I am not saying that a fully integrated theory of audio won't combine those elements, merely that I haven't seen it developed yet.
 
The following users thanked this post: IanMacdonald

Offline julianhigginson

  • Frequent Contributor
  • **
  • Posts: 783
  • Country: au
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #65 on: August 23, 2017, 01:36:19 am »
We already have plenty of DACs that do oversampling internally, so the nyquist filtering becomes cheap and not so phase mangling in the range that you care about.... It seems all they are doing is using a separate chip to oversample and then feed it into a DAC already oversampled?



I guess that gives them some control of the details of the interpolation, but I doubt it's going to be that different from what's being used by everyone else. As the action of this interpolation is happening well out of the range of human hearing, and what it generates only exists to allow a long filter transition band.

And as mentioned already they can scale up the bit depth all day long (and show dodgy graphs of sampling resolution error that look like they're representing about 4 bits and then maybe 6-7 bits of sample depth, not 16 bit) even if that little AKM part receives its bitstream at 32 bit depth, no way is an audio signal even approaching 18 bits of resolution and appropriate noise floor coming out the audio out pins on that little low power battery-run part... :-)
 

Offline ruairi

  • Frequent Contributor
  • **
  • Posts: 302
  • Country: us
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #66 on: August 23, 2017, 07:17:32 am »
While much audiophoolery is hogwash, the example I provided earlier is a case where I can't explain an audio result based on the best tools I have from science - as an electrical engineer with advanced degrees and years of experience using and applying frequency response, power, dV/dt and other elements which supposedly explain it all.  I am not saying that a fully integrated theory of audio won't combine those elements, merely that I haven't seen it developed yet.

The problem as I see it is that music is a type of magic and at it's finest it can transcend language, reality and reason to take us to another place.  In truly great music 1 + 1 can = 3!  Musicians and people who are sensitive to that magic want to believe that the same applies to the technology of music.  Engineers on the other hand are rational creatures to the end and refuse to leave room for the magic in how we interpret music. Simplistic notions of distortion and noise deny just how good we are at hearing subtle differences. 

I've spent over 20 years working professionally as a recording and mastering engineer and I get to work daily in an extraordinary studio that was built from the ground up to be as close to perfect as possible.  In the last 10 years I've been teaching myself electronics to answer the questions of why certain gear sounds the way it does and to try to figure out what makes a difference.

I've spent time in the room with some of the best listeners on the planet and I've also had the pleasure of spending time with some extraordinary design engineers.  Very few people in the world can do both at a high level.  That is my goal and I have a feeling it will take a lifetime.

 
 
The following users thanked this post: A Hellene, Alex Nikitin

Offline IanMacdonald

  • Frequent Contributor
  • **
  • Posts: 943
  • Country: gb
    • IWR Consultancy
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #67 on: August 23, 2017, 05:51:35 pm »
There was a time when FFTs first became available, when speaker manufacturers were using such tools to design speakers with near-flat frequency responses, something which had previously been near-impossible. Only thing was, they sounded truly dreadful.

Which, I think underlines the folly of chasing a specification parameter which has little bearing on actual results. In practice, audio gear need not have an absolutely flat response in order to sound good, and our ears actually adjust to the tonal balance of the sound anyway. A speaker which booms on bass or puts sibilance on vocals is never going to sound good though, and avoiding such artifacts is more important than a flat response. 
 
The following users thanked this post: A Hellene, eugene

Offline alexanderbrevig

  • Frequent Contributor
  • **
  • Posts: 700
  • Country: no
  • Musician, developer and EE hobbyist
    • alexanderbrevig.com
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #68 on: August 23, 2017, 08:35:38 pm »
A complete aside, and only really relevant to the above; IIR are much better for correcting speaker travel and it's response. At least in my limited experience :)

I'd also like to point out that even though psychoacoustics sounds like a pseudo science it really is not. It's a field which explains our perception of sound. Knowing the fact that we are more sensitive to baby screams than we are to rocks falling can be important to know. Much like the fact that we use echo and phase to locate and determine our surroundings. Then there is intermodulation and various noise shapes. What good is it that the RTA shows a flat line if you feel your ears bleed because of the 2-3k.

Every field of science turns into a belief or religion at some point. We're still or waiting to know what the higgs boson will tell us. Many people have dedicated their lives to 'sides' of that.

We guitar players love the power chord because it sounds deeper than it is. That is psychoacoustics, and it's true for everyone. Also, powerchords sohuld be played like the attached photo for extra psychology points ;) \../,

EDIT: Completely forgot to "answer" the question.
Yes you can make it sound better by shifting the bits, if you believe it will sound better.
« Last Edit: August 23, 2017, 08:37:59 pm by alexanderbrevig »
 

Offline IanMacdonald

  • Frequent Contributor
  • **
  • Posts: 943
  • Country: gb
    • IWR Consultancy
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #69 on: August 23, 2017, 09:37:49 pm »
My understanding is that the ear's 'microphone' is basically a long coiled tube with sensors along it at regular intervals. The standing waves set up in the tube determine which sensors are activated. This is why we perceive certain sets of frequencies as being a musical scale, and certain combinations of notes as working together, whilst others clash. In other words the concepts of an octave scale and chords are purely a  human standpoint. Other beings might hear our music very differently. Possibly not so much  :clap: as  :-//
« Last Edit: August 23, 2017, 09:41:31 pm by IanMacdonald »
 

Offline The Soulman

  • Frequent Contributor
  • **
  • Posts: 948
  • Country: nl
  • The sky is the limit!
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #70 on: August 23, 2017, 09:45:25 pm »
There was a time when FFTs first became available, when speaker manufacturers were using such tools to design speakers with near-flat frequency responses, something which had previously been near-impossible. Only thing was, they sounded truly dreadful.

Which, I think underlines the folly of chasing a specification parameter which has little bearing on actual results. In practice, audio gear need not have an absolutely flat response in order to sound good, and our ears actually adjust to the tonal balance of the sound anyway. A speaker which booms on bass or puts sibilance on vocals is never going to sound good though, and avoiding such artifacts is more important than a flat response.

Yes!  :-+

Together with:

I'd also like to point out that even though psychoacoustics sounds like a pseudo science it really is not. It's a field which explains our perception of sound. Knowing the fact that we are more sensitive to baby screams than we are to rocks falling can be important to know. Much like the fact that we use echo and phase to locate and determine our surroundings. Then there is intermodulation and various noise shapes. What good is it that the RTA shows a flat line if you feel your ears bleed because of the 2-3k.

Tells what is important in a high quality sound system: a optimized phase response and a frequency response that is free of resonant spikes (no you can't properly correct that with eq).

On topic: IMHO over-sampling is over-rated.

 

Offline alexanderbrevig

  • Frequent Contributor
  • **
  • Posts: 700
  • Country: no
  • Musician, developer and EE hobbyist
    • alexanderbrevig.com
Re: Can you truly make 16bit audio sound better by upping it to 32bit?
« Reply #71 on: August 23, 2017, 09:55:02 pm »
Tells what is important in a high quality sound system: a optimized phase response and a frequency response that is free of resonant spikes (no you can't properly correct that with eq).
Hehe, as a live audio engineer I know all about that. With analog gear, you almost always just moves the peak to another place when using the EQ. On my Midas I have phase linear EQ which helps, but a nasty room will stay nasty :(

Also, I have a bad habit when I'm drunk. I always sing a sine sweep in the loo to find the resonant peak and then do drop it like it's hot so it shakes the room (on that freq). I often get the sink for myself after ;) heh

On topic: IMHO over-sampling is over-rated.
Agreed. But if you want to send media over to f.ex a blue ray production. You will get strange looks/angry emails if you ship a 16@44.1kHz.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf