For crying in a bucket Frank can it be that difficult?
1. There is a real difference and if you measure after switch on and then 1 hour later you will see a difference
2. Or, you are acclimating to the sound period.
3. Or, there's some other rational explanation . . .
If anyone is curious this is where the concept of 0dBFS came from.
http://www.ieee.li/pdf/adc_evaluation_rf_expo_east_1987.pdf
http://www.ieee.li/pdf/adc_evaluation_rf_expo_east_1987.pdf
If you call for science you should ask yourself if the Meyer/Moran publication is up to the standards.
Although some additional information was given later on the website a lot of points remain unclear.
They did not analyze the content of the various discs.
They did not track the number of trials with each of the discs.
There was no measured data on the reproduction chain(s); the loudspeaker used for most of the trials seems to have a treble control (stereophile measured this loudspeaker model frequency response to be 8-10dB down at 20kHz for "flat" indication)
After a couple of months they noticed that the player used did have a problem with low level linearity, so they decided to use another one.
Did the player had this problem from the beginning? Unknown
Number of trials done while the player had this problem? Unknown
Total number of listeners is unknown. Number of trials for each of the four locations is unknown. Number of trials per participant is unkown (the article mentioned 10 trials but that did not meet the total number of trials; furthermore to choose 10 trials leads to very high probability of beta errors if the differences were not that large and/or the listeners did not have very high detection ability under test conditions)
No positive control was used, no calculation of effect sizes or statistical power was done, so to choose 10 trials is simply nonsense.
(Leventhal did explain that in 1986, and Jacob Cohen wrote his book about the importance of test power back in 1968 * )
At least one of the experimenters did express his opinion about the only possible result of such tests quite plain in publications of the boston audio society, so that might have had some impact on the club members participating.
(Even SY cites Rosenthal in his article...
) Nevertheless, the opinion of the authors may be correct, but their experiment was indeed (imho fundamentally) flawed.
* Jacob Cohen: Statistical power analysis for the behavioral sciences.
Last edited:
where are all the well designed, well executed listening studies with positive results on high res formats?
it seems a bit implausible that we wouldn't have robust positive evidence if home music playback listening discrimination of digital audio resolution beyond 16/44 were a 1 to 2 sigma human auditory discrimination ability
it seems a bit implausible that we wouldn't have robust positive evidence if home music playback listening discrimination of digital audio resolution beyond 16/44 were a 1 to 2 sigma human auditory discrimination ability
Last edited:
Up to now i don´t know about one, but did that affect the quality of the Meyer/Moran article?
Edit: At that point i am with SY, it depends on the question/hypothesis to be examined.
Mean estimate for human ability (worldwide?) is something completely different than "might be of importance for a small subgroup interested in high quality reproduction"
Last edited:
Well, Richard has implied that there is an AES paper/preprint/whatever by Floyd Toole that shows this. But try as I might, I haven't been able to manage to pry loose even the title of this thing.
I'm beginning to think that either it doesn't exist, or if it does, I'm going to read it and find that it doesn't say what Richard has been claiming it says.
I mean he also linked to a Sean Olive blog post that wasn't even germane to any of this. That blog post had to do with comparing 128kbs MP3 to CD for crying out loud.
And I have to say I've lost a lot of respect for Olive. He allowed himself be used for a Harman propaganda video that's a straight up fraud. He's supposed to be a serious researcher. That he allowed Harman to use him in such a way just shows that his integrity can be bought.
se
I think we agree on this point, but differ in the conclusions for simple tests.
We differ at that point.
It is at first quite impossible to do some "ears only tests" as you can´t switch off your brain.
Science knows a lot about the various bias mechanism at work and most of these are still working in a "blind test" .
Just some exerpts from that list were Rosenthal effect, Hawthorne effect, internal criterion problem. habituation error, presentation order effect and ... and...
Is there any argument why these bias mechanism were disabled if we only call a test "unscienitific" ?
Obviously not and therefore i was asking in which way the simple test should be done.
Simple or not, you have to decide in which way the "blinding" should be done (quite often double blind will not work at home, so it might be only single blind with the help of a trusty friend) and you have to decide how many trials you have to do.
And what about the number of positive results you need?
All these questions have to be addressed regardless of the "nonscientific" approach and all these points might have an influence on the results as long as the listener is not used to the specific experimental conditions.
I mean, although it will not be a strict scientific experiment you nevertheless will be interested in results that allow to draw conclusions about the difference and not about the ability of a listener to hear reliable and correct under unusual listening conditions.
Yes, sometimes it seems so. But obviously you have to be present when something like that happens before you´ll believe it.
Experience helps a lot; do some controlled tests with listeners with positive controls and see what happens.
Probability is high that you´ll notice a learning curve for most participants even with differences they hear in sighted tests.
In which miracolous way should your brain be able to suddenly avoid all the afore mentioned bias mechanism while processing the input of the auditory system?
See my answer to Tom above;
Quite often the difference in "air vibrations" is enough, but listeners fail to detect it in a controlled listening test....
Last edited:
I agree, you can't avoid those just like that. I agree that making a test 'blind' only removes part of the non-auditory influences, but not all.
In that regard a blind test is (in my view a lot) better than a sighted, non-controlled test, but it certainly is not perfect.
What would be the result of those remaining influences? Would they not lead to the tendency to hear a difference where there is none? What's your take on that?
Jan
The Distortion of Sound [Full Film] - YouTube
They confuse and conflate data compression and dynamic range compression. What really makes this video a complete fraud comes at 11:45 in the video where they do a graphic and audio comparison. Throughout the video they're going on about lossy data compression, i.e. MP3, with such gems as MP3 "removes 90% of the music."
But the comparison part is absolutely fraudulent. Just look at the waveform graphic for the "Compressed Audio." It's obvious that they're using gobs of DYNAMIC RANGE COMPRESSION on the track, AND they're dramatically reducing it in LEVEL. It also sounds like there was a bunch of clipping.
I've done MP3 encoding down to sub-96kbps and that's NOT what the waveform looks like and that's NOT what it sounds like. Looking at the waveforms in the same manner as in the video, the waveforms look no different.
It's a complete fraud and Olive should be ashamed of himself for taking any part in it.
EDIT: This was basically a propaganda piece for Harman's Clari-Fi, which purports to put back what lossy data compression takes out.
http://www.clarifisound.com
se
Last edited:
Steve,
I don't even need to watch the video or understand the science to know the difference between data compression and lossy encoding of any type pf data and dynamic range compression.
It is a shame that Sean Olive would give his name and reputation to something like that. At the same time we don't know if he was contractually required to do that, believe me that happens. He does seem to smart to have said those things but who knows the circumstances behind them.
ps. It seems the same thing goes on with the bit rate and frequency conversation that is going on here. There is a big difference between the usable dynamic range of a playback system and the dynamic compression that is taking place in most recordings. It really doesn't matter if you have a great system that can actually reproduce that range if you don't have the source material that can take advantage of it. I mean really except for the occasional orchestral recording or chamber music or that type of music how often isn't it still the case that the person doing the recording is not using a limiter compressor to cover their butt? Or they are riding the gain during the entire process to keep from clipping the signal at some point, and anyone who has done live sound knows you just can't be that fast, you can't do it instantaneously with signal, just not going to happen by human intervention. Let's face it almost 100% of all music is gain controlled in some manner before it is recorded at the end of the chain.
I don't even need to watch the video or understand the science to know the difference between data compression and lossy encoding of any type pf data and dynamic range compression.
It is a shame that Sean Olive would give his name and reputation to something like that. At the same time we don't know if he was contractually required to do that, believe me that happens. He does seem to smart to have said those things but who knows the circumstances behind them.
ps. It seems the same thing goes on with the bit rate and frequency conversation that is going on here. There is a big difference between the usable dynamic range of a playback system and the dynamic compression that is taking place in most recordings. It really doesn't matter if you have a great system that can actually reproduce that range if you don't have the source material that can take advantage of it. I mean really except for the occasional orchestral recording or chamber music or that type of music how often isn't it still the case that the person doing the recording is not using a limiter compressor to cover their butt? Or they are riding the gain during the entire process to keep from clipping the signal at some point, and anyone who has done live sound knows you just can't be that fast, you can't do it instantaneously with signal, just not going to happen by human intervention. Let's face it almost 100% of all music is gain controlled in some manner before it is recorded at the end of the chain.
Last edited:
It's number 3, there is an explanation, because there was a real difference, ie. also 1. But the measuring will be the killer challenge - I'm just saying that the conventional types of measuring won't pick it up, only to the degree of the normal variations one sees in looking at a number of competent, currently available pieces gear. Yet that difference was the essential element in making the system worthwhile listening to: prior to the change it wasn't, afterwards it was.
The point I was making is that this was an extreme example of something I hear all the time when listening to systems - that the subjective qualities alter over a lengthy periods of time because various parts of the system are constantly adjusting into different states of equilibrium, for all sorts of reasons. Most people don't seem to have a problem with mechanical behaviours causing variations over spans of 10's, 100's of minutes - the TT, cartridge "thing", for example - yet there seems to a major hangup as soon as electrical behaviours are mentioned, there appears to be a need to assert that the behaviours are locked into stone; once measured, at any time, that they are immutable ... sorry, it just ain't that way ...
With regard to blind testing, if the people are sensitive to the variations then the test setup needs no special preparation - the person listening can be outside the room, with zero visual awareness of the playback chain, or what's being done with it. I have no trouble picking when my playback is "in the zone" or not, when I totally away from the space - just like people can pick a real piano, or a "pretend" playback, before they even walk into a room ...
I'm sure you do, as well as most everyone in this thread. But you REALLY need to watch the demo at 11:45. It will make your blood boil.
Because not only did they use a bunch of dynamic range compression (because they needed to make the graphical waveform look different), but after the compression was applied, they greatly reduced the level. Because if they hadn't, the compression would have made it sound louder, and as a consequence likely "better" to the casual listener, which was their target audience. And I think they also used some hard limiting to cut off the peaks to make it sound obviously distorted as well.
Because if they HAD used an honest comparison, like say to a 128kbps MP3, the waveform graphics would have looked identical (at the scale they were displaying it at) and the vast majority of people wouldn't have heard any difference. So they had to resort to a deception that a blind person could see and a deaf person cloud hear.
I'd like to think that a contract requiring you to participate in, and give your tacit approval to a fraud wouldn't be an enforceable contract.
He was past president of the AES and still sits on a number of committees. I wonder if the AES would approve?
Oh sure. Compression is like any other tool and can even be used creatively to good effect.
se
Last edited:
Hi se.....i was looking for the paper.... I found the excerpt I used from Toole a few dozen lines back. But not from where i got it yet. Its in one of my many folders. But, dont worry, that isnt my main point. Here it is --->
This is sort of how my pea brain works -- conceptually most of the time. So I drew a diagram. BTW, only one person got it quit awhile ago;
View attachment bits vs thd vs dyn range.pdf
In the area of the dynamic range being played and listened to..... you have much fewer bits and higher distortion...... easily in the 0.1 to 1% for the bulk of the mid level music as was mentioned by another.
I was suggesting that is the reason CD doesnt sound good to many people like JC. In same way, if the 24 bit replaced the 16 bit as shown, the same level would be much lower distortion while listening. That is all there is to it IMO.
Good night -- way past my bed time and I have a painful tooth ache to deal with in the morning.
THx-RNMarsh
This is sort of how my pea brain works -- conceptually most of the time. So I drew a diagram. BTW, only one person got it quit awhile ago;
View attachment bits vs thd vs dyn range.pdf
In the area of the dynamic range being played and listened to..... you have much fewer bits and higher distortion...... easily in the 0.1 to 1% for the bulk of the mid level music as was mentioned by another.
I was suggesting that is the reason CD doesnt sound good to many people like JC. In same way, if the 24 bit replaced the 16 bit as shown, the same level would be much lower distortion while listening. That is all there is to it IMO.
Good night -- way past my bed time and I have a painful tooth ache to deal with in the morning.
THx-RNMarsh
Last edited:
With all respect, Richard, but I seriously doubt anyone could hear 0.1% distortion, unless something funny was going on in the harmonic decay rate. 1% perhaps, but we don't KNOW that, do we?
And unless we set some ground rules on audibility, we will never really know.
Some time ago, Nelson Pass said that he liked to let a little distortion in. In view of his work, rather well known and respected, I believe he was really saying why push it if it's already inaudible, except to get a "better" spec sheet. I hope he's reading this and can clairfy if I have misunderstood him.
On the other hand, there was a person here some time ago who claimed he could hear THD of just 0.001%, and I'd give real money to see him pull that one off (it's in the log somewhere).
And unless we set some ground rules on audibility, we will never really know.
Some time ago, Nelson Pass said that he liked to let a little distortion in. In view of his work, rather well known and respected, I believe he was really saying why push it if it's already inaudible, except to get a "better" spec sheet. I hope he's reading this and can clairfy if I have misunderstood him.
On the other hand, there was a person here some time ago who claimed he could hear THD of just 0.001%, and I'd give real money to see him pull that one off (it's in the log somewhere).
You're full of crap, Richard. There is no AES paper by Toole doing double blind testing of 16 bit digital and finding it lacking on the playback side.
I finally waded back through all this crap and pieced it together.
When I said that no one has demonstrated that 16 bit is lacking on the playback side versus 24 bit, you responded:
not true... gave sources already of F.Toole and dblt of 16bit.
I then asked you for the specific source.
You said:
Its an AES paper ...... the full gory details can be found in the References here:
x@#$%&&.... cant attach... again file size too large.....
I said I found 28 AES papers/preprints/whatever with Toole as one of the authors, and none of them appeared to have anything to do with 16 bit audio and double blind tests.
And this is the latest from you:
Hi se.....i was looking for the paper.... I found the excerpt I used from Toole a few dozen lines back. But not from where i got it yet. Its in one of my many folders.
So I went through your previous posts and this is what you had said:
It has been well reported that 16 bits didnt sound good enough. While back in the late 1990's Floyd Toole with usual thorough listening tests, said:
Here's where you first mention Toole and his "thorough listening tests" and had a short PDF snippet from some AES paper that says:
As Dr. Toole, one of the top experts in ACOUSTICS AND SPEAKER DESIGN is fond of saying, "Two ears and a brain are much more analytical than A MICROPHONE AND A METER!"
The clowns who wrote the AES paper are quoting Toole in a context that has NOTHING AT ALL TO DO with digital audio, let alone Toole doing any blind listening tests showing that 16 bits is inadequate!
Good god. You are a complete waste of time. You've not shown ANYTHING demonstrating that 16 bits are lacking on the PLAYBACK side while actually LISTENING TO MUSIC. Zip. Zilch. Nada.
se
- Status
- Not open for further replies.