Testing the claim: "I can hear differences between lossless formats."

Nov 15, 2014 at 10:47 PM Post #137 of 721
  What are the controls? Specify DAC, Amp, and of course HP, my own setup is OPPO BDP-105>Meier Classic>HD800 what are you going to set up, for me?

 
Oh, which equipment you use is up to you.
 
Do you know how to conduct and publish an ABX test with foobar2000? (That's the easiest way to go about this.)
 
I am also assuming you have already taken the 24-bit files and converted them to 16-bit.
 
Nov 15, 2014 at 10:56 PM Post #138 of 721
   
Oh, which equipment you use is up to you.
 
Do you know how to conduct and publish an ABX test with foobar2000? (That's the easiest way to go about this.)
 
I am also assuming you have already taken the 24-bit files and converted them to 16-bit.


Simple as pie with XLD, I do it all the time, for instance I was blowing leaves in my yard today and I wasn't using my AK 100ll, I was using my iPod and loaded up with 128 and 256 MP3's, no reason to do serious listening just background music which is fine.
 
Nov 15, 2014 at 11:17 PM Post #140 of 721
Interestingly, a few people have claimed to be able to reliably hear a difference between 24-bit and 16-bit audio as well.

@daltonljj
@Existence
@RUMAY408


 
You are hereby invited to strut your stuff and show us if you can demonstrate this with statistical significance!

(Or if you've done tests before, you can present documentation.)

How to explain it?
I can only tell the difference when using my laptop (HP Elitebook 2780p) and my AF180s, can't tell the difference on an iPhone (it can play FLAC through apps)
The sound just sounds fuller, richer than 16bit. Here's my tests on myself.

Playing Whitaker - My Own (44100hz, 16bit)
Then going to
Human Heat - Under My Skin (96000hz, 24bit)
The vocals on Under My Skin are extended, I can instantly tell its a 24 bit. Most 24 bit songs are really fatiguing for me to listen to for a long period of time, and I've always had very good hearing.

So, I threw myself at random flac files and guessed which was 16 and which was 24. About 75% I guessed right to 24bit. The way I can explain is is that 24bit has this extension in the sound I can hear that I can't hear in 16bit, like it's lacking something sonically. I caught myself trying to find this in my iPhone's music the other day, couldn't find it.

I use Audiofly AF180 IEMs unamped. I'm 16 and have very good hearing. I'd like to do audio tests scientifically one day, it'd be good to see if I'm not just BS'ing myself or this is real, but I can tell the difference between the two.
I use VLC, btw.
 
Nov 15, 2014 at 11:23 PM Post #141 of 721
   
Ah, so you use OS X. Do you have access to any Windows PCs? (foobar2000 won't normally work on Apple systems.)


Sorry the Apple MacAir is my only PC.
 
For most HP's I honestly don't think I could tell a difference between 320 or 256 MP3's and Hi Res 24/96 or above.  
 
The higher end HP's, properly amped, and with an above average DAC, and 24 bit makes a difference to me, no further claims at all.
 
If you want to run a double blind placebo controlled study using controls that account for age groups, music sources, and every HP, DAC, and Amp, I would love to see it and that includes a sample size that is large enough to also account for audio interest and indifference. 
 
Nov 15, 2014 at 11:27 PM Post #142 of 721
How to explain it?

We'll worry about explaining it later. First, you need to prove you can actually hear a difference and this isn't just placebo.
 
Download foobar2000 if you don't have it yet, then download this ABX Comparator plugin. Pick a 24-bit file you're confident you can differentiate, then convert that file to 16 bits using foobar's built-in converter (make sure to select 16-bit output and dither). You may need to install the FLAC frontend to convert to FLAC, but you can use WAV instead.
 
Then follow these instructions to load both files in the comparator, make sure volume is matched, and test away. Try to do 15 to 20 trials, the more you do the easier it will be to pass with confidence. Post your results when you're done!
 
Making the 16-bit file yourself from the 24-bit file is important to avoid using two different masters, which will sound different but not because of the bit depth. Doing a test like this is very important to avoid sighted bias.
 
 
Sorry the Apple MacAir is my only PC.
 
For most HP's I honestly don't think I could tell a difference between 320 or 256 MP3's and Hi Res 24/96 or above.  
 
The higher end HP's, properly amped, and with an above average DAC, and 24 bit makes a difference to me, no further claims at all.
 
If you want to run a double blind placebo controlled study using controls that account for age groups, music sources, and every HP, DAC, and Amp, I would love to see it and that includes a sample size that is large enough to also account for audio interest and indifference. 


Not much point conducting a test of every headphone in existence until someone is able to prove they can hear the differences they claim they hear at all.
 
Nov 15, 2014 at 11:28 PM Post #143 of 721

 
That is not a scientific test at all; it's just casual listening, to different songs, no less.
 
First and foremost, you absolutely need to compare the same master of one song at a time - specifically, the 24-bit version of the file and a 16-bit version converted from that file. Secondly, it needs to be done in a methodical fashion. Members here can guide you on how to do this. If you are ready to do so immediately, I can also instruct you how to conduct an ABX test.
 
The reason some 24-bit files sound different is not because they are 24-bit per se, but because they were derived from a different master. So in your listening so far, you're not really comparing 24-bit and 16-bit so much as you are comparing how those recordings were mastered in a general sense.
 
For reference, I will link you to this article. (It's shared a lot, I know, but just in case you haven't seen it yet.)
 
Nov 15, 2014 at 11:36 PM Post #144 of 721
  Sorry the Apple MacAir is my only PC.
 
For most HP's I honestly don't think I could tell a difference between 320 or 256 MP3's and Hi Res 24/96 or above.  
 
The higher end HP's, properly amped, and with an above average DAC, and 24 bit makes a difference to me, no further claims at all.
 
If you want to run a double blind placebo controlled study using controls that account for age groups, music sources, and every HP, DAC, and Amp, I would love to see it and that includes a sample size that is large enough to also account for audio interest and indifference. 

 
Nothing quite so ambitious. I just wanted to investigate certain claims, for example, if you took some 24-bit files, converted them to 16-bit, and perceived differences between how they sound. That would involve testing those files on your system, specifically.
 
Here and here are two software programs that you can run some tests with on OS X. However, I have no direct experience with them, so you will have to work with the provided instructions.
 
As a general rule, you want to run 15 or 20 trials (like Head Injury said), record your results, and share them here.
 
Nov 16, 2014 at 12:14 AM Post #146 of 721
First and foremost, I really don't know what your intention of this post is. Because you sure come off very offensive as if you're trying to dis people of their beliefs be it real or placebo. Not a very nice thing to do in a forum imo. 
 
Anyway back to your topic does 24bits matter? if so prove how you can test it? A least that is what i understand you asking me.
 
To tackle the problem I will first ask you. Assuming all thing equal. Same DAC  and same media player. If you were to play the same file (both in 16 and 24bits or even if 32 bits if you're so anal about it) and use like some device compare their analog signal wave forms when played. Is there a difference? The answer is gonna be a yes and also depending on the DAC that you use this difference will vary.
 
As of how to prove humans can hear it I would have to say that IMHO there is no way to scientifically test and prove that I can hear it. Reason is that there are no 2 human beings in the world that are identical. They way we perceive sound is different. Take for an example. Having a flat freq response. What may seem flat to one may sound bass heavy to another.The perception of sound is attributed to a lot of factors. On the human it could be the shape, size of your year canal and even the sensitivity of the individual's hearing. Another thing is that the difference with are talking about here is very minute. So an analogy that I can come up with is if I were to give you 2 identical balls, one 1kg another 2kg can you tell the weight difference? Obviously you can else something is really wrong with you. Now same experiment. with mass 2kg and 2.1kg can you fell the difference most probably still yes. Guess what some ppl might tell you they don't feel the difference. They start to dis you and ask you to prove how you can feel that 10% change in weight when you are not even a weighing scale. Point i'm trying to put across here is that human sensitivity varies too. So frankly I think its hard to scientifically prove that human's can tell the difference between 16bit and 24bit.
 
Or how about this prove to me how you can test that humans can't hear the difference between 16 and 24 bits.
 
Thus to sum things up it matters but whether you can hear it is a different thing all together. 
 
Nov 16, 2014 at 12:30 AM Post #147 of 721

 
My intention was not to offend, but to scientifically investigate whether people who claim to can actually hear a difference between playback of lossless formats, whether it's WAV vs ALAC or 16-bit vs 24-bit.
 
If you claim to hear a difference, you should be able to demonstrate this with statistical significance via a series of tests, such as the ABX tests mentioned throughout this thread and others. If you fail the tests, it means you were merely guessing and can't actually hear a difference. But the first step is to determine whether you are willing to conduct these tests in the first place.
 
If science offends you, perhaps it would be best to avoid the sound science forum. Some of the more experienced members here are less polite than I. If you're fine with the pursuit of science in relation to sound reproduction, I suggest reading through this thread to get a better idea of what is going on in digital audio.
 
Nov 16, 2014 at 12:33 AM Post #148 of 721
Just a note about using Windows laptops.  Some have soundcards built in that have to be reset for the resolution.  You may have the headphone output set to 44/16, and then when you listen to 96/24 though capable of 16 it will on the fly convert the 24 to 16 bit and the sample rate to 44 khz.  The quality of this on the fly conversion varies and is sometimes audible even vs a 16 bit original. 
 
Usually 24 bit originals are at least 48 khz or 96 khz.  Pick let us say 96 khz and 24 bit originals.  Convert the 96/24 to 44/16 using good software.  Foobar with the Sox plugin or Audacity version 2.0.3 or newer (which also uses Sox).  Convert the 96/24 to 44/16 and then convert 44/16 back to a second 96/24.  The resolution of the original will be reduced to 44/16 but you can compare both in Foobar.  If you only care about 24 vs 16 bit, then simply convert 96/24 to 96/16 then back to 96/24 and compare. 
 
Nov 16, 2014 at 12:35 AM Post #149 of 721
  First and foremost, I really don't know what your intention of this post is. Because you sure come off very offensive as if you're trying to dis people of their beliefs be it real or placebo. Not a very nice thing to do in a forum imo. 
 
Anyway back to your topic does 24bits matter? if so prove how you can test it? A least that is what i understand you asking me.
 
To tackle the problem I will first ask you. Assuming all thing equal. Same DAC  and same media player. If you were to play the same file (both in 16 and 24bits or even if 32 bits if you're so anal about it) and use like compare their analog signal wave forms. Is there a difference? The answer is gonna be a yes and also depending on the DAC that you use this difference will vary.
 
As of how to prove humans can hear it I would have to say that IMHO there is no way to scientifically test and prove that I can hear it. Reason is that there are no 2 human beings in the world that are identical. They way we perceive sound is different. Take for an example. Having a flat freq response. What may seem flat to one may sound bass heavy to another.The perception of sound is attributed to a lot of factors. On the human it could be the shape, size of your year canal and even the sensitivity of the individual's hearing. Another thing is that the difference with are talking about here is very minute. So an analogy that I can come up with is if I were to give you 2 identical balls, one 1kg another 2kg can you tell the weight difference? Obviously you can else something is really wrong with you. Now same experiment. with mass 2kg and 2.1kg can you fell the difference most probably still yes. Guess what some ppl might tell you they don't feel the difference. They start to dis you and ask you to prove how you can feel that 10% change in weight when you are not even a weighing scale. Point i'm trying to put across here is that human sensitivity varies too. So frankly I think its hard to scientifically prove that human's can tell the difference between 16bit and 24bit.
 
Or how about this prove to me how you can test that humans can't hear the difference between 16 and 24 bits.
 
Thus to sum things up it matters but whether you can hear it is a different thing all together. 


There's actually no significant differences in the waveform, nothing humans can hear at any reasonable volume. The difference between 16 and 24 bits will be in the guesswork being done to recreate the analog signal. Rounding errors result in quantization noise, the more rounding the more noise. 16 bit already offers us a noise floor of -96 dB. We only need more bits if we can hear that noise 96 dB below the signal above the ambient volume in our room, or the noise of our electronics. We can't, in any realistic situation. And even if we could, no piece of equipment can play 24 bits at full quality (24 bits offers -144 dB, the absolute best SNR I know of for an amp or DAC is 130 dB).
 
Here is a picture of a quick null I did between a 24 bit file and the 16 bit file I made out of it. This is, basically, the difference between the files.
 

 
 
It's very easy to prove humans can hear the difference. Just pass an ABX test with 95% confidence as I just outlined a few posts ago. No complex and horribly misleading analogies needed, either you can reliably tell the difference blind or you can't.
 
In order to prove humans absolutely can't hear the difference, we'd need to test every human on the planet until we found someone who passes. In order to prove humans can, at least occasionally, hear the difference, all you need to do is pass. It's a lot easier if you lend a hand, don't you think?
 

Users who are viewing this thread

Back
Top