so any "analog" is going to show up a ADC/DAC pass through?
so in your formulation - no difference from a live mic feed comparison, or tape played back on the same machine, or tape played back decades later on different model, different heads, unknown alignment differences? how about ips, track width?
while my skepticism earlier was with respect to vinyl playback and ADC/DAC, digital RIAA
you're certain these subtle distinctions can be heard even at the inner diameter?
not allowing for the expected distinctions, limitations of "analog source" dosen't help credibility
isn't this moving the goalpost, over generalizing?
so in your formulation - no difference from a live mic feed comparison, or tape played back on the same machine, or tape played back decades later on different model, different heads, unknown alignment differences? how about ips, track width?
while my skepticism earlier was with respect to vinyl playback and ADC/DAC, digital RIAA
you're certain these subtle distinctions can be heard even at the inner diameter?
not allowing for the expected distinctions, limitations of "analog source" dosen't help credibility
isn't this moving the goalpost, over generalizing?
Last edited:
We have to fight our tendency to prove our self right.
Please think also of the possibility that an ABX or whatever blind test may provide the opportunity to filter out our own biases, leaving the digital problems intact.
George
No you have not. You just spout unsubstantiated anecdotes about this.
There's only one of your 'associates' or 'competitors' who, contrary to you, ever got the guts to put his money where his mouth is, and he failed the 'apparent'.
You seem to, as usual, be creating fake news here John.
Boston Audio Society - ABX Testing article
Jan
It is not just me who hears digital changes, many others do as well, and they are the people I care about when it comes to audio quality.
Bringing up IVOR's double-blind test from 30 years ago or so, shows that Ivor was a bit overconfident, and he got his 'tail wacked'. I would not fall for such a test, as I know the 'results' already in advance. Yet, you still bring up this one test, over and over.
Those of you who believe that digital is almost impossible to note, especially over time, can live with it, for all that I care.
Bringing up IVOR's double-blind test from 30 years ago or so, shows that Ivor was a bit overconfident, and he got his 'tail wacked'. I would not fall for such a test, as I know the 'results' already in advance. Yet, you still bring up this one test, over and over.
Those of you who believe that digital is almost impossible to note, especially over time, can live with it, for all that I care.
you're suggesting SACD is "good" for digital audio today is a laugh itself, John
we're pretty sure changing your opinion ain't happening either - but some of are committed to change if provided evidence to a reasonable scientific standard required by all other fields of human endeavor today
you're "side" has also had those 30 years to come up with positive DBT moving human hearing resolution estimates, ample opportunity to publish
we're pretty sure changing your opinion ain't happening either - but some of are committed to change if provided evidence to a reasonable scientific standard required by all other fields of human endeavor today
you're "side" has also had those 30 years to come up with positive DBT moving human hearing resolution estimates, ample opportunity to publish
Last edited:
Of course I 'cop out' of such a test EVERY TIME! I know that the test is inherently flawed to detecting many small errors, not all of them of course, but the important ones still left to improve on. I first mentioned this about 38 years ago in the 'Audio Amateur'. Nothing much has changed, in my opinion the ABX test and its derivatives is flawed!
Tell me John, just how is the ABX test flawed? My own experiences with that procedure shows me that it's great at getting directly to the answer without interference from your expectations. The best way to benefit from those tests is to do your best at listening and accept the results however they turn out.
Normal listening tests are also good as long as your subjects don't have any clue as to what to expect. That's in cases were differences can be large. They are best when comparing one device to another where both are prototypes. That being a truly blind listening test.
-Chris
Normal listening tests are also good as long as your subjects don't have any clue as to what to expect. That's in cases were differences can be large. They are best when comparing one device to another where both are prototypes. That being a truly blind listening test.
-Chris
Well this is going to go nowhere. Everyone's coming from their perspective with dug in heels. Oh well...
Jan,
When the tester who is familiar with the device under test has no issue determining when it is in circuit or out, and the test subject cannot then the conclusion is not that the device is perfect rather that the test subject is the issue. (Additional confirmation as to hearing differences when none existed.)
The conclusion should be there were issues affecting the test and they should be eliminated to further examine the issue.
Of course not. The original issue was the insertion of a digital device in a live microphone system. This should be quite obvious under certain conditions having nothing to do accurate signal transmission. (There is a time delay when the digital device is inserted... Duh!)
Experiment design is a critical issue. We just did a simple A/B test here to see if a newly repaired loudspeaker matched its' mate. The conclusion was it did not and the mate requires a woofer recone also. (Single blind the listeners did not know which was which but the operator did.)
Last edited:
When the tester who is familiar with the device under test has no issue determining when it is in circuit or out with a Blinding protocol in place...
Infamous 2007 Atkinson interview of Gordon Holt:
45 Years of Stereophile | Stereophile.com[/QUOTE]
Infamous 2007 Atkinson interview of Gordon Holt:
45 Years of Stereophile | Stereophile.com[/QUOTE]
Last edited:
I thought your comment was broader
I'm not not still stuck on the 30 year old test with a "leak" in the Blinding - no reason to rehash - just get it right
I'm not not still stuck on the 30 year old test with a "leak" in the Blinding - no reason to rehash - just get it right
If you do a traditional ABX test, you may get null results. If you do the expect same equipment swap but don't tell the wife, she tell you it sounds different.
The ABX is flawed in the way it's conducted, not in the merits.
The ABX is flawed in the way it's conducted, not in the merits.
The reference I made to an experiment with a mic was simply a statement of what was done. It was not intended to imply a mic is in any way equivalent to a phonograph record as a source.
Also, it was not my intent to say that inserting data converters in an analog signal path would necessarily be the worst source of distortion in the signal chain.
The point I was trying to make was that it looks like a miniDSP-HD may be roughly similar in specs to a Lynx2 card. My listening experience with a Lynx2 is far more extensive than the one experiment I described. Based on that experience, I don't think I would want to have something like that in a high quality analog signal path if I could avoid it. Not that it would necessarily be the worst thing in that path. But if someone had a particularly nice record, perhaps of a beautiful and very carefully recorded symphony, why add more audible distortion in the playback path if it could be avoided?
That's the direction I was trying to go in by posting the two wav files. They are only 16/44, yet the distortion of the Lynx2 A/D is distinctly audible. The Crane Song HEDD sounds much better. If I was going to insert data converters in my analog path, I would probably be okay with the HEDD, but not so much with the Lynx2.
Now, it may be that a miniDSP-HD is closer to a HEDD/DAC-1 data conversion path than a Lynx2 data conversion path. I don't know, as I haven't had an opportunity to try a miniDSP-HD. However, the miniDSP-HD test article Bill pointed to has and IMD graph that looks to me like there might be some audible distortion, maybe small, but roughly down around where quantizing noise from undithered 16/44 would be.
So, although I haven't heard one of those boxes (and I would like to), I decided to raise a question as to whether or not its distortion contribution would be desirable in an otherwise very high quality phonograph system. My position would be that I would definitely want to listen to one before assuming from looking at specs that inserting such a box in my analog signal path would be of negligible consequences.
As an aside: I wonder if anybody has had a chance to see if they can hear any difference between the files. https://www.dropbox.com/sh/9r5kiiptr00seub/AAC5zEynkg-ZmriNSkQxFPaEa?dl=0
Regarding ABX testing, I like the idea and it seems like it should be workable in many cases. However, I don't think the testing should be conducted in an adversarial atmosphere. Friendly, relaxed exploration into what can be heard and verified with ABX testing would be more likely to leave test subjects in a more receptive frame of mind for hearing small differences without distraction.
That being said, for hearing very small differences in distortion or other aspects of sound, statistical methods would probably be needed. It would seem reasonable to expect that down around the threshold of discrimination, some false positives and false negatives would occur. If somebody can be successful even 70% of the time, given sufficient trials to get reliable statistics, that would be significantly different than 50% of the time.
In addition, I think the ABX test system should be carefully vetted to make sure that it's possible to hear very small differences on it. One way to test the whole system would be to test with increasing smaller amounts of distortion with human listeners, or perhaps a mic could be set up in front of the speakers/headphones to accurately measure the presence of the distortion. Its not clear to me that ABX playback systems used in some prior research were fully capable of accurately rendering very small distortion signals. Also, if using speakers, I think test subject needs to be in the very near field, as room reflections and HF attenuation with distance both tend to obscure small subtitles of sound.
That being said, for hearing very small differences in distortion or other aspects of sound, statistical methods would probably be needed. It would seem reasonable to expect that down around the threshold of discrimination, some false positives and false negatives would occur. If somebody can be successful even 70% of the time, given sufficient trials to get reliable statistics, that would be significantly different than 50% of the time.
In addition, I think the ABX test system should be carefully vetted to make sure that it's possible to hear very small differences on it. One way to test the whole system would be to test with increasing smaller amounts of distortion with human listeners, or perhaps a mic could be set up in front of the speakers/headphones to accurately measure the presence of the distortion. Its not clear to me that ABX playback systems used in some prior research were fully capable of accurately rendering very small distortion signals. Also, if using speakers, I think test subject needs to be in the very near field, as room reflections and HF attenuation with distance both tend to obscure small subtitles of sound.
Last edited:
If you are referring to the 32 tone test that's a pretty brutal scenario and even then the hash is well below the level of, say the Sousa test and would certainly be swamped by the distortion from the cartridge which can run several %.
I certainly did fret at first of the evils of putting and AD/DA chain in place, but engaging my rational brain put that at rest. For me, right now, this is the right solution.
This is of course not to say that miniDSP doesn't have issues but they are known and can be worked around. For example the lack of analog level control and a slightly higher than SOTA noise floor does mean you have to have to be careful with gain structure or put a multichannel preamp after it. Jan designed a board for the Behringer DCX2496 to address this, but the kit for that costs as much as the DCX does to start with, so was outside of my budget. Hypex do the DLCP which is also nice and if I were doing this again I might go that way. If you don't mind having a PC in the system you can go the whole hog and use ultimate equaliser Bodzio Software . That is a step to far for me. If I ever get a study again then maybe...
And at the end of the day it's fun, which is really the point of all of this 🙂
I certainly did fret at first of the evils of putting and AD/DA chain in place, but engaging my rational brain put that at rest. For me, right now, this is the right solution.
This is of course not to say that miniDSP doesn't have issues but they are known and can be worked around. For example the lack of analog level control and a slightly higher than SOTA noise floor does mean you have to have to be careful with gain structure or put a multichannel preamp after it. Jan designed a board for the Behringer DCX2496 to address this, but the kit for that costs as much as the DCX does to start with, so was outside of my budget. Hypex do the DLCP which is also nice and if I were doing this again I might go that way. If you don't mind having a PC in the system you can go the whole hog and use ultimate equaliser Bodzio Software . That is a step to far for me. If I ever get a study again then maybe...
And at the end of the day it's fun, which is really the point of all of this 🙂
This, more than anything else is the difficult part. And it has more to do with the human than the equipment.
The thing about equalizers is that we tend to focus on the frequency domain, and neglect transient response graphs and waterfall plots. In addition, its well known that we humans can adapt to the frequency response of a system more than we can adapt to transient response problems. That's one of the reasons people still use NS-10s to mix, its not because of flat frequency response, clearly. But one can hear things on them that are very difficult to hear on other speakers. So, it would seem that if we want to hear everything in our source material, we might want to give more attention to things in addition to frequency response.
- Status
- Not open for further replies.