Gentlemen, and any ladies...
We have several co-existant things (variables) floating about here. It's important to be aware of them, and take them into account.
- measured "distortion" does NOT correlate to "subjective fidelity."
- a source of "extra" euphonic coloration MAY cause a system that has "extra" artifacts of very low level but higher order harmonics to "sound right." (even if the system specs at "low distortion")
- the whole thing is a fairly large compromise to begin with, so the expectation of more is not realistic.
- ...something else that escapes me at the moment that is important, so it will have to wait to be mentioned later...
My beef with DBTs that I am aware of (fyi, SY, and others) is that no one seems to be nailing down these "confounding" factors when publishing their test results - sure they are "correct" but only for the specific test conditions, which may or may not have various flaws that are undocumented.
_-_-bear
We have several co-existant things (variables) floating about here. It's important to be aware of them, and take them into account.
- measured "distortion" does NOT correlate to "subjective fidelity."
- a source of "extra" euphonic coloration MAY cause a system that has "extra" artifacts of very low level but higher order harmonics to "sound right." (even if the system specs at "low distortion")
- the whole thing is a fairly large compromise to begin with, so the expectation of more is not realistic.
- ...something else that escapes me at the moment that is important, so it will have to wait to be mentioned later...
My beef with DBTs that I am aware of (fyi, SY, and others) is that no one seems to be nailing down these "confounding" factors when publishing their test results - sure they are "correct" but only for the specific test conditions, which may or may not have various flaws that are undocumented.
_-_-bear
Funny when I said an expert luthier, a professional gutarist, and a music instructor at Berkely could not tell the difference in an open loop FET and TL-082 as a guitar pickup, I was told that musicians make bad listeners. BTW they all were given instruments blind and practiced/played at their own comfort and leisure.
I agree that there may or may not be confounding factors; nothing in life is perfect, and DBT's aren't either. BUT, most DBT's do carefully list the circumstances and conditions, and try to keep it as objectively (controlled) as possible and try to use more than a single listener to get a statistically relevant result.
That is much, much more that a subjective, 'I sat down, plunket up a CD and I found that xx sounded muchg better than yy'-type of totally uncontrolled event.
To put more trust in such an anecdotal event than in a reasonably well set up DBT is only logical if you are unwilling to give up your believe.
Which is your prerogative, of course, but let's then recognise it for what it is.
jd
That's because you ended up at the wrong conclusion, silly. If a test gives you the answer you don't want, blame the test. First rule of high end audio.
Jan,
The usefulness of the published DBTs is something I question. No doubt that they may be of value. However, I have yet to read a published DBT that actually printed the results of measured and in-situ tested equipment. And by that I mean more than just the IM and THD of the amplifier(s) - although that would be a start. I think I've made this complaint before - the ones that i have read are insufficiently documented such that one could actually replicate the test conditions to any degree of reliability or certainty.
Put this another way - I have heard perfectly "good" systems that use gear that is presumably of "low distortion" and "good engineering" that to me personally sound like "dog poop", meaning that I couldn't discern one thing from another. A DBT test on such a system would be statistically valid, but imo quite meaningless beyond that particular system.
I've also heard systems that are actually "ok" but they were so horribly set up that again, I found it impossible to really hear enough of what was going on so that I would be comfortable making any determination as to a "component" being one way or the other...
I still like the IROC/NASCAR analogy - all the right parts set up just a bit wrong and the thing slams into the wall on the curve...
_-_-bear
The usefulness of the published DBTs is something I question. No doubt that they may be of value. However, I have yet to read a published DBT that actually printed the results of measured and in-situ tested equipment. And by that I mean more than just the IM and THD of the amplifier(s) - although that would be a start. I think I've made this complaint before - the ones that i have read are insufficiently documented such that one could actually replicate the test conditions to any degree of reliability or certainty.
Put this another way - I have heard perfectly "good" systems that use gear that is presumably of "low distortion" and "good engineering" that to me personally sound like "dog poop", meaning that I couldn't discern one thing from another. A DBT test on such a system would be statistically valid, but imo quite meaningless beyond that particular system.
I've also heard systems that are actually "ok" but they were so horribly set up that again, I found it impossible to really hear enough of what was going on so that I would be comfortable making any determination as to a "component" being one way or the other...
I still like the IROC/NASCAR analogy - all the right parts set up just a bit wrong and the thing slams into the wall on the curve...
_-_-bear
Scott,
Your anecdote is not surprising at all.
There are two main problems with "professional musicians" and sound.
1. Some have lost a significant part of their hearing as the result of being over-exposed to amplified music.
2. Most musicians who have significant training "hear" the music in their heads - not in the "air". This is especially true of those who can sight read music... it is a problem in some situations.
_-_-bear
Your anecdote is not surprising at all.
There are two main problems with "professional musicians" and sound.
1. Some have lost a significant part of their hearing as the result of being over-exposed to amplified music.
2. Most musicians who have significant training "hear" the music in their heads - not in the "air". This is especially true of those who can sight read music... it is a problem in some situations.
_-_-bear
In a linear network for a perfect step you get 100% of the settling on final value in 100% of the time allowed. An RC network settles to 63.7% of the value in one time constant (R*C = 1 time constant). It then settles to 63.7% of the remainder in the next time constant. (76.9% of final value) Of course it will never actually reach the final value, so custom is to allow 5 or 7 time constants. Since you want fast settling higher order filters are use.
So if your timing accuracy is really bad say 1% then a linear system will have a 1% level error, the inverse exponent system having dropped faster at first will have less error. With zero timing error the linear system should have no error and the inverse exponent must have some.
So if all your steps are single bit both methods are nice, but when you do a step of 1/2 full scale linear should still be zero error and inverse exponent will have the single bit remainder multiplied by the step size.
Jan,
The usefulness of the published DBTs is something I question. No doubt that they may be of value. However, I have yet to read a published DBT that actually printed the results of measured and in-situ tested equipment. And by that I mean more than just the IM and THD of the amplifier(s) - although that would be a start. I think I've made this complaint before - the ones that i have read are insufficiently documented such that one could actually replicate the test conditions to any degree of reliability or certainty.
Put this another way - I have heard perfectly "good" systems that use gear that is presumably of "low distortion" and "good engineering" that to me personally sound like "dog poop", meaning that I couldn't discern one thing from another. A DBT test on such a system would be statistically valid, but imo quite meaningless beyond that particular system.
I've also heard systems that are actually "ok" but they were so horribly set up that again, I found it impossible to really hear enough of what was going on so that I would be comfortable making any determination as to a "component" being one way or the other...
I still like the IROC/NASCAR analogy - all the right parts set up just a bit wrong and the thing slams into the wall on the curve...
_-_-bear
Bear, I'm sure you can find fault at anything you want if you look good enough (and I don't mean 'you' personally, but in general). My point is that published DBT's, with all their flaws and shortcomings, real or not, are infinitely more reliable than a casual, subjective, non-documented test with a system that's totally in the dark. Except, of course, for YOU.
Compare for instance that well-documented test that Ivor Tiefenbrunn flunket with a 'test' done by me or you sitting at home and plugging an ADC-DAC in or out of our system and then telling all the world that I/you clearly heard a difference, that the DAC-ADC "destroyed the realism" or whatever. Is it unreasonable to have more trust in my/your results than in the Tiefenbrunn test results, except if you are unwilling yo give up your believe. Which is my/your prerogative, but let's recognise it for what it is.
jd
D-right, meet my half Polish cousin, deaf as a bat after 30y in the business and still at it.
=> YouTube - Brainbox - Down Man
Jan,
I do not know the details (and by the details I mean more than a description and picture of the room) of this flunked test by some person of "standing".
This is exactly my point. Not knowing enough of the empirically determinable details puts this test or ANY test into question.
Surely, you can not suggest that (to exaggerate for clarity) a DBT performed with a mid-fi receiver and run-of-the-mill 3-way speaker as the central components will be likely to yield useful information as to something like two high-end DACs?? (assuming no gross flaws in either one?) Clearly, the specifics of the test conditions is crucial, not casual.
What some other person said or did not say is a red herring in this case. He might be right, he might be wrong. He's got an hypothetical construct, that's all. And, of course a belief.
What i am saying is that a DBT is fine as a tool, but it is only as good as the test conditions permit - and the test subjects. If the test conditions are not sufficiently well documented and tested then it is merely a statistical reflection of a specific test condition and circumstance and MAY OR MAY NOT be generalizable to ALL cases...
Let's recall we are not talking about what is "good enough for the average person" we are talking about what is possible to achieve at the limits of the current technology!
_-_-bear
I do not know the details (and by the details I mean more than a description and picture of the room) of this flunked test by some person of "standing".
This is exactly my point. Not knowing enough of the empirically determinable details puts this test or ANY test into question.
Surely, you can not suggest that (to exaggerate for clarity) a DBT performed with a mid-fi receiver and run-of-the-mill 3-way speaker as the central components will be likely to yield useful information as to something like two high-end DACs?? (assuming no gross flaws in either one?) Clearly, the specifics of the test conditions is crucial, not casual.
What some other person said or did not say is a red herring in this case. He might be right, he might be wrong. He's got an hypothetical construct, that's all. And, of course a belief.
What i am saying is that a DBT is fine as a tool, but it is only as good as the test conditions permit - and the test subjects. If the test conditions are not sufficiently well documented and tested then it is merely a statistical reflection of a specific test condition and circumstance and MAY OR MAY NOT be generalizable to ALL cases...
Let's recall we are not talking about what is "good enough for the average person" we are talking about what is possible to achieve at the limits of the current technology!
_-_-bear
Well you could read up on the test about the conditions, equipment, the pre-testing, the checks and balances, it's all there, well documented. I don't want to become personal, really not, but the fact that you personally don't know the details is because you didn't look them up. That's hardly something you can blame the test or the testers for.
Of course that casual test we did in our homes with our equipment is totally in the dark, so trusting that is equivalent to buying a used car after a phone call to a used car salesman
.jd
Jan,
Please feel free to send me via email or link the best documented "DBT" you have? I'd be quite pleased to correct my comment in your favor, if it is warranted. I said I have yet to read one that I found to be even marginally well documented. So, if I have not seen it, why not just send it along?
SY, c'mon... I never said anything about "peeking at the nameplate"!! Your reading comprehension is off - that is the opposite of what I said. I'd expect far more detailed and comprehensive empirical and scientific measurements and tests.
As far as what Rod said, I dunno... don't think I made a direct value judgement on what he said rather I commented on the overall issue...
_-_-bear
Please feel free to send me via email or link the best documented "DBT" you have? I'd be quite pleased to correct my comment in your favor, if it is warranted. I said I have yet to read one that I found to be even marginally well documented. So, if I have not seen it, why not just send it along?
SY, c'mon... I never said anything about "peeking at the nameplate"!! Your reading comprehension is off - that is the opposite of what I said. I'd expect far more detailed and comprehensive empirical and scientific measurements and tests.
As far as what Rod said, I dunno... don't think I made a direct value judgement on what he said rather I commented on the overall issue...
_-_-bear
I do not look at nameplates. I just keep track of the components under test. I don't have to be told 'which is which'. I just need to know that this is L and this is M for example, or D and H. This nameplate thing is a 'slander' to compromise my experience, in making intelligent and correct audio decisions.
Its obscure to me that higher order filters necessarily provide faster settling. As far as I can see, the settling time isn't so much a function of the filter order as the Q - a higher order bessel filter will settle quicker than a somewhat lower order elliptic for example.
I don't consider a level error (linear) of 1% to be particularly worrisome. Do you? The faster the settling, the smaller this level error. 5 time constants will give an error -43dB down.
Sorry, I'm not following here. Both methods?
I don't particularly keep track of all tests I see, but the one I brought up before is here:
Boston Audio Society - ABX Testing article
There's an interesting one on speaker relative ranking blind vs sighted by Floyd Toole here:
http://www.harmanaudio.com/pv_obj_c...0EFDC694000200/filename/audio_art_science.pdf
(the whole paper is a good read but the blind/sighted part is at this chapter "BLIND vs. SIGHTED TESTS – SEEING IS BELIEVING")
Now I am sure you can and will find flaw with either of those papers, but remember: even a serious attempt at documenting and controlling tests runs rings around the usual sighted, casual anecdotal 'tests'.
jd
- Status
- Not open for further replies.