How many people can hear the difference between...

spitzer

New member
...any two supposedly equivalent audio related things, really?

No, I am serious. This came up again the other day, as a friend thought one of my recordings could've done with a bit more "presence" etc. etc. I know this is an age-old topic in itself, but I did spend time and could not find anything close to a simple answer, and a simple answer to THIS should actually exist and I'd even expect it to be easily found but that's not the case.

The things being compared could be something different, but let's say "A" is a piece of music at CD quality and "B" is the same piece of music compressed to an MP3 at "maximum" quality. It could also be things at different sample rates or whatever, that's not the point.

How many people, in pure numbers, x people out of y people, can reliably tell the difference between "A" and "B"? What is the proportion of people who are actually capable of telling them apart?

There's a ton of very heated debate on this stuff, but this is something I haven't found anywhere. I've seen various studies comparing musicians, engineers, "audiophiles" etc. to each other but the groups of people themselves are "floating" and unspecified. Where did they get those people? What are they supposed to represent? None of them answer the most relevant question (to me). This type of question is interesting in the general population, not at all that much otherwise if you ask me (it turns into a contest instead...)

So, anyone here know of a good study or two?
 
One of those "generates more heat than light" topics. Thanks for bringing it up :).

I can't point to a study, but I am certain that the number of people that can objectively and reliably identify original vs. lossy (not just saying they hear a difference) is very small, and it depends, of course, on a hearing ability well above the median, very good sound equipment and the kind of source material that makes the compression algorithm [decisions] more apparent.

The other question I don't see asked is whether any study can show that there's a significant population that can say their life is diminished because of the use of lossy compression. Jeez, talk about first-world problems.
 
Spitzer, I think its possible that your friend might have a point, especially if his hearing acuity is significantly different from yours.

I've seen some smaller tests where people had trouble distinguishing between 320K mp3s vs the original CD file. One in particular used musicians who, you would think, would be attuned to such things as timbre and clarity.

I have, however, heard a few instances where a particular passage gave MP3 encoding trouble, but they were very minor, and it was years ago. I doubt that I could do the same today, as my hearing had degraded over the past 10-15 years.

One the other hand, Keith makes a point about how much it really matters. I worked in printing for over 40 years, and part of the job involved color accuracy. While I could throw 20 identical pages and tell variations in the tone, contrast and shading, I often thought "but does it make a difference if the jeans are slightly greenish blue vs the original reddish blue? As long as the price is right the customer is going to buy it!" Of course I NEVER said that to the customer. They expected it to be a match, and that's what we tried to achieve.
I find sound to be much more difficult to correlate than with vision, as it tends to be more transient in nature. You can study an image and focus on impurities or inaccuracies, but with sound recording, it is constantly changing. Its meaningless to "freeze" a single note and study it.

With art (and creating music IS art) there are no absolutes. Its about perception which is a personal issue. What you might find perfectly acceptable, someone else might find dull and lifeless because they listen for something different. The cymbals might not have the sparkle they expect, or they might be overbearing.
 
One of those "generates more heat than light" topics. Thanks for bringing it up :).

Yes. I should stress that I'm not, I'm really not, wanting to start the typical debate. The specific point here, like I said, is: why is the actually relevant phenomenon (whether or not people in general, not "audiophiles" or professional audio engineers can tell) NOT being studied but almost everything beside it is? I mean, for example it's very common knowledge that older people start losing perception of higher frequencies, happens to EVERYONE regardless of general health. So after age 50, what's the point of sampling frequencies that even you yourself at half your age could not possibly hear? (just an example! please. No ill will meant.)

(and yes I did see that smiley face. :) Just honestly looking for actual "real life" data which I can't seem to find.)
 
In a strictly controlled A/B blind comparison most people would detect a difference. However, my experience is that some people prefer the sound of the mp3. I don't know if it's the compression or the other artefacts but mp3s can be nicer to listen to and perhaps even less tiring. When it comes to playing one after the others = in maybe an iTunes stream with different formats and quality levels, very often I have noticed people just don't notice, if the Robbie Williams .wav is followed by the low bandwidth mp3 at 160, or the more bits/higher sampling rate wav that follows if they like the material. It's always subjective and rarely objective because it's music, and your liking or disliking of the song colours your judgement. objectively we can measure and analyse but listening isn't like that. Noise and distortion get spotted, but I firmly believe once above that, format is just a matter of taste - like years ago when people would record on cassette with Dolby B, the replay with it off because they liked the extra HF and could ignoramus's the hiss!
 
Spitzer, I think its possible that your friend might have a point, especially if his hearing acuity is significantly different from yours.

In this particular case the big, the huuuuge if is that... well not technically an 'if' I guess. Anyway, not sure about hearing acuity, his (available, even if not actual) listening equipment is much better though... but he had not heard the original tracks! Never even once, let alone solo'ed, compared or anything like that. Just off-hand, he suggested a different mic placement to get "more of the cymbals (high end) in there", which to me was odd for at least two reasons: 1. I thought there were too MUCH of the cymbals (mid-high), they were bulldozing the guitars 2. I did nothing deliberate whatsoever to alter the freq response, no EQ at all (quick mix). So, what I might very well have missed would still all be in there. 3. What I heard with my cheap "gear", speakers and phones both, very closely matched the cymbals natural sound. They just happen to be fairly dark and don't have that much of the very high-end harmonics.

This was all just what got me thinking about the broader topic a bit more, almost like he heard something missing that was never there. Or didn't hear something he would have liked to hear. Certainly a testament to people hearing the same stuff differently.

You've all made some very good points here. How much does it matter, there are no absolutes. Still looking for an actual study or even any kind of test though. The one point where I would definitely and strongly, without proof, personally disagree is

rob aylestone said:
... most people would detect a difference. ...

That is assuming you're still referring to CD vs. best possible mp3. You might consider this nitpicking, but still. "Most people" means at the very least more than half of ALL people.

And still, the most interesting thing would be to see any study of this kind in the general population. Let's say the sample population would be people ages 15-84, not legally deaf. Which way would it go?
 
I find it depends on what you are listening to the music on, and how you are listening to the music.

If you playback system is very high end you will notice some difference. If you are listening to bud type earphones, computer speakers, etc. no difference.
If you have the music playing at home or in the car and you are not really listening its just music for company, no difference. If you are absolutely listening to the song, every note every expression (for example classical music, Jazz Etc) then maybe you will notice.

Alan.
 
My concert pianist friend can detect micro tuning defects on hearing somebody play the piano, he has perfect pitch and comes out with crazy things. Dampers need looking at on the F sharp above middle C, or Ab 7 is a tiny bit high. He listens to all our tracks and then comes back to me with editing fixes. He listens very, very carefully. Usually I send wavs, but occasionally I mistakenly send him the crunched MP3 files I edit for the samples on the website. He has never, ever noticed. So his very good hearing misses completely what we are talking about, but picks up totally different things, but clearly these tiny levels of detail are in the music. What exactly are we hearing when we listen for quality differences?
 
Don't know if this adds anything. At 65 (and still recording, about to release a new 'album') my hearing has deteriorated to the extent that I am deemed in need of hearing aid in left ear for "marginal" hearing loss.
Talking to my "hearing therapist" (aint the NHS wonderful), she claimed that I perceived my loss of hearing more acutely BECAUSE I'm a long time musician recording engineer.
The standard NHS test for whether one needs hearing aid help apparently lies 15db (in your ability to hear) to either side of the NHS selected frequency (which I do not recall) (Hope that makes sense.
So someone could hear at 15 db ABOVE the chosen volume when that frequency is played and STILL get help, or be 15db below.
Sorry: Here's my point. I can tell the difference in a vocal if I add 1 or 2 db without my hearing aid, and when I told her this, she told me about musician's ability.

SO "ordinary folk" are what the NHS tests target and they nbeed a 15 db difference whereas, theoretically I need a 1 tgo 5 db difference.
I don't think ordiunary people can truly tell the difference at all and that they don't care either.

PS: I have a new fangled hearing aid with a special setting for lisening to music.
 
The NHS hearing test is appallingly bland. The actual frequencies are 5 or 6 across the range and therefore incapable of drawing any kind of response curve that is meaningful to people used to talking about sound.

If you do your own hearing test with a tone generator in your studio, you won't of course have headphones that are completely flat - but they'll be gentle and only plus/minus a few dB. by watching the meters and being honest, you can build up a very accurate curve showing your own hearing - and even allowing for the frequency response of your headphones, your hearing loss at specific frequencies will be quite evident. I suppose you could even use a microphone with very flat response to build up an inverse EQ curve to flatten your test headphones to increase the test effectiveness, but I'm positive I would get more accurate curve than the NHS one can produce, because it is ONLY concerned with speech. The private audiologist my mother visits explained their equipment can do much more narrow band testing, but this is not deemed necessary for speech. If you pay for the test, you can have third octave testing done - the machines are quite capable of it.
 
How many people, in pure numbers, x people out of y people, can reliably tell the difference between "A" and "B"? What is the proportion of people who are actually capable of telling them apart?

There's a ton of very heated debate on this stuff, but this is something I haven't found anywhere. I've seen various studies comparing musicians, engineers, "audiophiles" etc. to each other but the groups of people themselves are "floating" and unspecified. Where did they get those people? What are they supposed to represent? None of them answer the most relevant question (to me). This type of question is interesting in the general population, not at all that much otherwise if you ask me (it turns into a contest instead...)

So, anyone here know of a good study or two?

Why does this even matter...if there was/is some "study"?
It will not prove anything, because it is impossible to test every person on the planet...so the study will not be accurate. To take some sample average and churn out some "proof", is no more accurate than doing a political poll and hope it's right. ;)

I get the feeling that what you really want to know is why your friend hears it differently than you...and right there is your answer...people do.
Women differently from men, in general, children different from adults, in general, younger different from older, in general...and that's not even getting into the subjective aspect of WHAT they are hearing...that's through audiology testing, that it's been shown to be the cases, in general.

I also really, REALLY hate the boring argument that some will make when they say "most people won't even hear it or notice it, so who cares"...etc..etc.
If you can hear it...if one other person can hear it...then assume you/they are not the only ones...so don't approach any audio production by catering to the lowest common denominator of the general public...but rather go with the assumption that some will hear it, and if it needs correcting, then do it.

Of course...if it's just a purely subjective thing...you like it less bright and someone else likes it more bright...that's an artistic decision, and also, that's why most playback system have some level of EQ adjustment capability so people can tailor their listening experience to their tastes.
One thing IS and has been true with audio productions...if they are well balanced and have some room to "bend" without losing their soul...more people will be able to make personal taste adjustments when listening, without drastic alterations to your production.
 
Why does this even matter...if there was/is some "study"?
It will not prove anything, because it is impossible to test every person on the planet...so the study will not be accurate. To take some sample average and churn out some "proof", is no more accurate than doing a political poll and hope it's right. ;)

I get the feeling that what you really want to know is why your friend hears it differently than you...

You misunderstood horribly. This has nothing to do with how or why he or person X or person Y hears that particular track. Also, the NUMBER of people being tested is not the problem, the problem is WHO is being tested. Not regular people, but engineers, musicians and "audiophiles".

What I really want to know, I specified in the first post. Specified in broad terms but still specified.
 
You misunderstood horribly.

No I didn't.
It sounds to me like you're trying to understand why people hear a difference and who those people are...narrowed down to industry people.

Like I said, what does it matter if there was a study and even if it was done with engineers, musicians and "audiophiles...it's always going to be very subjective.
There have been countless tests done with the A/B of mics or monitors, and all kinds of other things...by industry people...and while many of them are about guessing which is A or B...right or wrong pick, most people always claimed to hear a difference. Which is my point.

What is it that you are really asking or trying to understand...stemming from your friend hearing something differently from you?
Let's assume there was a specific study...with some specific numbers...60% could hear a difference, 40% couldn't...or whatever.
What does that information do for you...what question(s) would it answer for you?
 
Last edited:
I have to agree with Miroslav for a number of reasons. There are many of our senses that defy objective testing for so many reasons, and will remain subjective for ever, I suspect. I didn't think much about this till I studied to be a teacher, with music and performing arts as my key area. We had to do loads of research and presentation tasks, and my teachers always used my ones to destroy (pleasantly) my firmly held opinions and beliefs, simply because they could not be validated and were always subjective. One of the people on the programme was a whiskey blender - and we viewed his specialist subject as completely subjective, as it was all about if you like alcohol which I don't, being allergic to it!

However, he had the results of tasting panels, chemical analyses and his results were very well received as facts. Mine were far less solid.

The problem is that our only qualitative ability to measure is in terms of frequency and amplitude. We can measure this very accurately. Oddly, we can measure distortion, and we can see it on screens in terms of waveforms. We accept evidence of audio files by reference to a small number of objective responses that measure and define very well - but as soon as we move to 'quality', objective response goes out of the window. It transfers to terms that frankly really don't make any sense. All musicians on this forum would listen to a Les Paul played through a decent tube amp, with some well known effects and processing and say this sounds better than the student guitar with the cheapest pedal in the shop. I suspect if we did a poll, we'd all say the same thing - one is better quality than the other. Very few, if any, could say why! Measurements would fail. Good distortion vs bad distortion? First thing is you need to define what distortion is. We'd probably go for differences between the original and the processed. How is the distortion different? Some might start looking at the differences between the harmonics - numbered as they get higher and measures and others might concentrate on evener odd numbered harmonics - all that stuff. Trouble is, with all the available tools to measure and analyse, without listening to the end result, could you actually look at the data and say - ah, this one will sound best? No - absolutely not. We listen to the track and then we validate our choice. You cannot do it the other way around. My teacher was a bit of a 70s synth fan, and he proved my analysis to be rubbish by bringing in a synth and an oscilloscope. He then told me to look at the screen and put 5 preset waveforms into order from horrible to wonderful sounding. Totally impossible. Result = subjective.

We all have some ability to determine quality, but it relies on conditioning to put what we hear into boxes. At some point our brain heard something and put it into a pigeon hole we then use to form judgements. It isn't just sound - two other sense have the same issue. Smell and colour. Colour is a particularly fascinating one to study. It's a subject where no two people can be proven to see the same thing. When you were a child, your parents pointed at the sky and said "blue". We then use this for our entire life. See that colour - say blue. If you were in fact seeing yellow, does it matter if you think that colour is blue? Could this be the answer to favourite colours and why we all have different ones? Some people cannot even describe colour - Purple, mauve, magenta, pink, fuchsia. See the problem. you can measure the wavelength of light but you cannot say with any certainty that what people actually see!

Any form of qualitative analysis of audio is confined to a very narrow accuracy range. Trying to go further forces it into objective subjectiveness, or subjective objectiveness and we cannot go further until somebody can measure what data crunching happens in our brains.
 
Miroslav knows "what I really want to know" better than me? And you agree with him? Honestly, are you crazy?

I'll address the other stuff later in case there's something there to address.
 
We had to do loads of research and presentation tasks, and my teachers always used my ones to destroy (pleasantly) my firmly held opinions and beliefs, simply because they could not be validated and were always subjective. One of the people on the programme was a whiskey blender - and we viewed his specialist subject as completely subjective, as it was all about if you like alcohol which I don't, being allergic to it!

However, he had the results of tasting panels, chemical analyses and his results were very well received as facts. Mine were far less solid.
...

Any form of qualitative analysis of audio is confined to a very narrow accuracy range. Trying to go further forces it into objective subjectiveness, or subjective objectiveness and we cannot go further until somebody can measure what data crunching happens in our brains.

I made no claims concerning what is subjective or objective, or what is not. I made no claims at all. I did not propose a plan to "go further" either.

Whatever problem with "objective subjectiveness" etc. you think would be in the tests... IS in the tests. Somebody is probably doing an A/B test right now, comparing lossless files against OGG/AAC/MP3 or whatever and is planning on publishing it somewhere. These tests are still being done all the time, with precisely the same problems you think they have. So... are you saying these studies should not be done at all?

The test methods being used are a different topic, what I wished for were tests where the people being tested were properly selected and specified.

What Miroslav started with is borderline "nothing at all should be tested or at least can not be trusted" and that is a subject I won't touch and is out of place here.
 
Streaming services including Google for youtube have done all these tests and they will not publish results. Perception is a huge part interpretation, plenty of studies showing that people can be fooled into thinking one sample sounds better than the other even when they are exactly the same just by telling them one is a higher bit rate sample even when it's not. Those that have done this kind of testing consider it proprietary and will tell you if you want an answer do a study yourself.

Food for thought YouTube

and YouTube
 
Miroslav knows "what I really want to know" better than me? And you agree with him? Honestly, are you crazy?

I'll address the other stuff later in case there's something there to address.

I asked you twice to get to the heart of the matter, because I really DON'T know what you really want to know...because it can't be just "have there been any tests".

You haven't replied.
 
Back
Top