Who's afraid of DBTs

[Jud]

Yes, we are both saying exactly the same thing - I'm talking about a single listening trial result & jud is saying that we don't know what number of the trials in an overall null result are because of "false negatives".

 

Yes, nicely said. Another way of saying the same thing - mmerrill is describing the actual mechanisms that might cause false negatives, while I described how you might go about determining if DBT was causing them to occur and how often.

[mmerrill99]

So, any thoughts on the DBT internal controls I outlined in post 364? Are they workable? Are they sufficient? What about practicalities in using them - do people view they would be easy/difficult to implement? Any other suggestions?

 

Should we start a different thread for this?

[PeterSt]

Should we start a different thread for this?

 

Say I'm fresh enough to think No. That is, if I think I know what you mean.

But that is my opinion.

[Jud]

Sorry just tired. And you already have the most informative posts here anyway. But since you got nada from scholar (presumably after trying hard) and I also got nada, we may have to go the wild net route.

 

And also, I am not doing any formal science. Just trying to get a feel on those FNs/FPs. Never really tried cause I dont think they are necessary. Audio is well known and well studied science. All available devices are built on ~100 years old theories like electromagnetism and waves. And it's not life & death. So I dont really see why would that kind of triple-sure assurance is needed.

 

Some people just dont want to accept that 99,99% is more than enough for all practical purposes and spend all their energy looking for the mythical 100%. Maybe nobody told them that 100% hardly exists outside the math class.

 

But anyway, I'm all for looking. Bring those FNs.

 

Yes, this gets to something important:

 

DBT for audio is a perfectly fine thing. If you want to say "I will not consider X to be true or a report of X to be reliable unless confirmed by DBT," good and well. The only thing that is *not* perfectly fine is saying "I rely on DBT, therefore I am on the side of Science, and all those who do not are anti-Science." IMO, we don't know yet whether use of DBT in audio is truly at a level of reliability that can be called scientific, so any individual's choice to rely on it or not shouldn't be taken as an indication of a pro- or anti-scientific attitude.

 

There are also other ways of advancing knowledge in the audio field that have excellent reliability. One such is finding measurable physical phenomena that correspond to what one hears (for example, Dennis's story about trying out a particular amp with his speakers, hearing the amp "strain," to use his term, and realizing it was due to a bad impedance mismatch, IIRC). Reliance on this way of advancing knowledge rather than DBT should, again, not result in someone being characterized as being anti-scientific. In fact, I think this is one way that many professional engineers in the field of audio work.

[mmerrill99]

And also, I am not doing any formal science. Just trying to get a feel on those FNs/FPs. Never really tried cause I dont think they are necessary. Audio is well known and well studied science. All available devices are built on ~100 years old theories like electromagnetism and waves. And it's not life & death. So I dont really see why would that kind of triple-sure assurance is needed.

Trithio, some of your posts are great because they bring together into one post, all the standard, stock answers from the DBT proselytiser's camp.

 

So we have the statements "Audio is well known and well studied science" - so what, there's nothing new to be discovered?

 

"100 years old theories like electromagnetism and waves." - so what - we are examining auditory perception & it's testing, not electronic engineering.

 

"So I dont really see why would that kind of triple-sure assurance is needed." What triple a-sure assurance are you talking about? I wouldn't call determining the error rate embedded in DBT results is a triple-sure assurance

Some people just dont want to accept that 99,99% is more than enough for all practical purposes and spend all their energy looking for the mythical 100%. Maybe nobody told them that 100% hardly exists outside the math class.

Now some of the wilder assertions - DBTs are 99% correct? Isn't that the very point of the discussion? - neither jud nor I accept this assertion without knowing the level of FNs in the results - something that has not been demonstrated. You already have made up your mind on the results as being 99% correct without the necessary evidence & this is why the DBTers are said to operate on a "belief system"

 

Thank you for the summarising of the thought process of those who fear examining DBTs as a suitable test, as usually administered outside of research labs.

[esldude]

Brass band and Lindsey Buckingham shouting at you? (Speaking of the title song, not the entire album.) No thanks. Van Morrison for me back in college days. Back when my wife and I were first dating I was going through a Peruvian flute phase, but thank goodness I moved on pretty quickly or I doubt she'd have stuck around.

 

Hold on now. I didn't say I liked it. I considered it the worst they had done when I first heard it. Just bad music. I still consider it the worst the group has ever done as an album. I do not have a copy nor want one. That however has no bearing if the gal you are with likes its, and likes it alot. That they spent like 2 years and big money to produce it only made it worse.

[esldude]

Concerning the DBT topic. I guess I would ask, since it clearly works quite well even in audio to a point what is better?

 

I agree with those saying overly simple DBT's are of limited value (but not no value). A big one is putting in positive and negative controls.

 

I have seen one circuit designer who had an interesting approach. If he was making a pre-amp using an op-amp he wondered how audible it was. So he set up to do a simple ABX test. His feeling was ABX might have a false negative problem and wanted to be conservative. He got a null with one op-amp. He cascaded two, got a null. And on and on until he got to 7 series op amps he planned on using. As his design would incorporate two in a row he felt since 7 were needed to get a barely positive result then using two was likely not a problem. I seem to recall he had to go to 11 or 13 until it was so bad he got 100%. Doesn't address everyone's concerns, but it is one approach of being conservative that is useful in some situations. In some areas of concern such an approach isn't possible.

 

I do like the type up-down tests that will determine the threshold of each individual listener. Those however are more difficult to arrange. Plus you need to have something you know you can make audible and reduce in steps for it to be workable.

[Jud]

Hold on now. I didn't say I liked it. I considered it the worst they had done when I first heard it. Just bad music. I still consider it the worst the group has ever done as an album. I do not have a copy nor want one. That however has no bearing if the gal you are with likes its, and likes it alot. That they spent like 2 years and big money to produce it only made it worse.

 

The Peruvian flute stuff was actually the background music for the first PBS Nature series, "Flight of the Condor:" The Flight Of The Condor Soundtrack (1985)

[Jud]

Concerning the DBT topic. I guess I would ask, since it clearly works quite well even in audio to a point what is better?

 

I agree with those saying overly simple DBT's are of limited value (but not no value). A big one is putting in positive and negative controls.

 

I have seen one circuit designer who had an interesting approach. If he was making a pre-amp using an op-amp he wondered how audible it was. So he set up to do a simple ABX test. His feeling was ABX might have a false negative problem and wanted to be conservative. He got a null with one op-amp. He cascaded two, got a null. And on and on until he got to 7 series op amps he planned on using. As his design would incorporate two in a row he felt since 7 were needed to get a barely positive result then using two was likely not a problem. I seem to recall he had to go to 11 or 13 until it was so bad he got 100%. Doesn't address everyone's concerns, but it is one approach of being conservative that is useful in some situations. In some areas of concern such an approach isn't possible.

 

I do like the type up-down tests that will determine the threshold of each individual listener. Those however are more difficult to arrange. Plus you need to have something you know you can make audible and reduce in steps for it to be workable.

 

Mike Moffat did specifically mention op-amps in his Head-Fi post that I quoted here.

 

Now as to whether his story regarding not being able to hear differences between amps with DBT but being able to hear them with longer listening sessions is true, I dunno.

 

It is interesting that regarding amps, using somewhat different processes, we both seem to like Spectral (by no means the cheapest amps - though I saved quite a lot on mine buying it well used - but by no means in the stratospheric range either). What are you using these days?

[Daudio]

I know many are afraid of HA .

 

Again you know wrong. Many CA folks just don't like HA, some have had bad experiences there, but afraid ? Hell no. And what is it with you and fear ? What are you afraid of so much that you accuse others of it ?

 

 

BTW, I think that you would be very happy over on HA. You should join up, and spend your time there, instead of feeling insulted and complaining about complaining. So give it a try. What have you got to lose ?

 

 

(I know what we have to lose, and I'm looking forward to it)

[Daudio]

The Peruvian flute stuff was actually the background music for the first PBS Nature series, "Flight of the Condor:" The Flight Of The Condor Soundtrack (1985)

 

Jud,

 

Ah ! I had that album until fairly recently, when I digitized a couple of my favorite tracks, and then sold it along with most of my vinyl, during my project to digitize and eliminate most of my physical recordings (still ongoing - the last 10-20%).

 

Saw the series, loved the music, bought the record

 

P.S 'El Condor Pasa' is good too

[mmerrill99]

Concerning the DBT topic. I guess I would ask, since it clearly works quite well even in audio to a point what is better?

But is this not what is being asked here - how well is it working? We don't have any internal self-checking mechanisms that can give us a handle on this.

 

I agree with those saying overly simple DBT's are of limited value (but not no value). A big one is putting in positive and negative controls.

I'm not sure what value you can put on a test whose results you are unsure of? By "A big one" do you mean that you consider using internal controls would be a big improvement?

 

I have seen one circuit designer who had an interesting approach. If he was making a pre-amp using an op-amp he wondered how audible it was. So he set up to do a simple ABX test. His feeling was ABX might have a false negative problem and wanted to be conservative. He got a null with one op-amp. He cascaded two, got a null. And on and on until he got to 7 series op amps he planned on using. As his design would incorporate two in a row he felt since 7 were needed to get a barely positive result then using two was likely not a problem. I seem to recall he had to go to 11 or 13 until it was so bad he got 100%. Doesn't address everyone's concerns, but it is one approach of being conservative that is useful in some situations. In some areas of concern such an approach isn't possible.

But, if as is being queried here, the ABX test is suppressing the perception of real audible differences then, in this light, might this example not be interpreted quiet differently?

[sandyk]

But is this not what is being asked here - how well is it working? We don't have any internal self-checking mechanisms that can give us a handle on this.

 

I'm not sure what value you can put on a test whose results you are unsure of? By "A big one" do you mean that you consider using internal controls would be a big improvement?

 

But, if as is being queried here, the ABX test is suppressing the perception of real audible differences then, in this light, might this example not be interpreted quiet differently?

 

I would be wary of any ABX that is performed using something like Foobar 2K as many here appear to do.. Foobar 2K is certainly not the gold standard for playback software, although for many it may be good enough to identify more obvious differences. An E.E. friend of mine has designed a remote controlled comparator box with relay switching that can be used to switch between various components. They are probably available commercially too, or as DIY projects.

[kumakuma]

The Peruvian flute stuff was actually the background music for the first PBS Nature series, "Flight of the Condor:" The Flight Of The Condor Soundtrack (1985)

 

Missed that one but I enjoyed Flight of the Conchords a lot.

[Superdad]

Yes, I think the sound on Jazz is great. I've just been looking at my vinyl LP versions and copyright date for Jazz was 1978, and Bop 'Til You Drop' was 1979. If Bop 'Til You Drop was the first digital one, then it must mean that Jazz is analogue.

 

I remember having to change Bop 'Til You Drop about four times before a got a flat one finally, and it still sounded poor compared with Ry Cooder's previous LPs.

 

Hi guys: I'm just catching up again.

With regards to Ry Cooder's "Jazz," I can assure you that despite "Bop 'Til You Drop" getting the publicity as Warner's first "all-digital" release, portions of "Jazz" were also done digitally by Lee Herschberg. I can't find links or references to prove it, but my very first job in audio was 1976 in 6th grade at Joseph Byrd Productions' home recording studio--dubbing commercial jingle demos on giant mono Presto machines and building a patch bay--for Joseph Byrd, the eccentric composer who did the arraignments for "Jazz." (This was also my first exposure to high-quality herbal 'buds'.) Anyway, my close school buddy at the time--a few years my senior--was also the recording assistant on "Jazz" (his only other credit was later on the score to Koyaanisqatsi). And I remember him telling me about the digital recorders they were fighting with in the sessions.

I think the levels ended up low on the LP because they were not used to the deep bass (some from the studio floor) coming off the new digital recorders and were afraid of cutter overload.

(As an odd, "it's a small world" aside, decades later my friend move to Finland and became an executive with Nokia--right around the same time that I think our own Miska was an engineer there.)

 

Okay, you can all go back to raging about DBT and what does or does not prove what many of us see and hear so clearly.

[esldude]

But is this not what is being asked here - how well is it working? We don't have any internal self-checking mechanisms that can give us a handle on this.

 

Sure we do. We have other ways beyond the DBT. Like nerve activity, structure of the ear itself. The amount of bandwidth and several other functions of the ear have DBT results in the vicinity those other parts of knowledge would predict. So we aren't wildly off. There is university research about how closely spaced we can hear tones and similar things. Monitoring of auditory nerve signals show just about the results DBT's have shown.

 

I'm not sure what value you can put on a test whose results you are unsure of? By "A big one" do you mean that you consider using internal controls would be a big improvement?

 

Well many tests are not at all black and white. I may have a multimeter that is +/- 10%. For many purposes that is good enough. More accuracy often would not matter. Blind testing is much like that. Don't trip up on the simpler parts and it may not be the final, final word, but the word it gives is far from no use.

 

But, if as is being queried here, the ABX test is suppressing the perception of real audible differences then, in this light, might this example not be interpreted quiet differently?

 

Depends upon the level of perception doesn't it. That designer can't be absolutely 100% certain his procedure didn't leave something the test didn't reveal. He does know the test revealed something with enough circuits seriesed together. He knows whatever it is doing to alter the signal using only two of the circuits the effect is far smaller. He knows that whatever difference there is will not be a truly large obvious difference. Seems like a good conservative approach likely to be successful.

[sandyk]

Okay, you can all go back to raging about DBT and what does or does not prove what many of us see and hear so clearly. - Alex C.

 

 

+1

[Jud]

Sure we do. We have other ways beyond the DBT. Like nerve activity, structure of the ear itself. The amount of bandwidth and several other functions of the ear have DBT results in the vicinity those other parts of knowledge would predict. So we aren't wildly off. There is university research about how closely spaced we can hear tones and similar things. Monitoring of auditory nerve signals show just about the results DBT's have shown.

 

 

Where are the hair/nerve cells sensitive to jitter?

 

I have little or no problem with the results of DBTs concerning loudness and frequency. I think variations in those sonic attributes are things that most people can identify pretty well without training.

[esldude]

Where are the hair/nerve cells sensitive to jitter?

 

I have little or no problem with the results of DBTs concerning loudness and frequency. I think variations in those sonic attributes are things that most people can identify pretty well without training.

 

The hair and nerve cells respond to the signal created by jitter or it simply would not be heard at all. Don't get sucked in to the idea super tiny bits of jitter are heard. Those that have been demonstrated as audible are pretty high levels that cause tones above the noise floor not related to signal causing them. So there is a known signal there to be perceived.

[sandyk]

Monitoring of auditory nerve signals show just about the results DBT's have shown.

 

Source ?

Perhaps monitoring of auditory nerve signals is unreliable at this point in time too ?

[sandyk]

Those that have been demonstrated as audible are pretty high levels that cause tones above the noise floor not related to signal causing them.

 

Some types of jitter are reported to result in low level wideband noise.

[mmerrill99]

Sure we do. We have other ways beyond the DBT. Like nerve activity, structure of the ear itself. The amount of bandwidth and several other functions of the ear have DBT results in the vicinity those other parts of knowledge would predict. So we aren't wildly off. There is university research about how closely spaced we can hear tones and similar things. Monitoring of auditory nerve signals show just about the results DBT's have shown.

 

Well many tests are not at all black and white. I may have a multimeter that is +/- 10%. For many purposes that is good enough. More accuracy often would not matter. Blind testing is much like that. Don't trip up on the simpler parts and it may not be the final, final word, but the word it gives is far from no use.

The point is you don't know the error rate of your results so you can't equate it to your multimeter - this has been calibrated, your DBTs have not

 

Depends upon the level of perception doesn't it. That designer can't be absolutely 100% certain his procedure didn't leave something the test didn't reveal. He does know the test revealed something with enough circuits seriesed together. He knows whatever it is doing to alter the signal using only two of the circuits the effect is far smaller. He knows that whatever difference there is will not be a truly large obvious difference. Seems like a good conservative approach likely to be successful.

As a designer I wouldn't want to trust a test system that might be suppressing audible differences as a means of determining if 2 opamps in series were audible - it's the equivalent of him using his multimeter but not knowing it's accuracy - I doubt any sane designer would do such a thing

[sandyk]

As a designer I wouldn't want to trust a test system that might be suppressing audible differences as a means of determining if 2 opamps in series were audible - it's the equivalent of him using his multimeter but not knowing it's accuracy - I doubt any sane designer would do such a thing

 

Years ago I had a David Tilbrook designed Preamplifier that used 2 NE5534AN opamps in tandem , but with an auxiliary output after the first one as well. Despite them being used in their 1,000s in recording Studios, the output directly from the first opamp sounded quite noticeably better to both myself and a friend.2 opamps in series may in some configurations result in a better S/N when the gain is split between them, but may not sound better.

Alex

[Allan F]

Though I could be wrong, I am fairly sure that Ry Cooder's "Bop 'Til You Drop" was the first digitally mastered album publicly released. The music on that album is wonderful.

 

+1

 

On both counts.

[trithio]

Trithio, some of your posts are great because they bring together into one post, all the standard, stock answers from the DBT proselytiser's camp.

 

So we have the statements "Audio is well known and well studied science" - so what, there's nothing new to be discovered?

 

"100 years old theories like electromagnetism and waves." - so what - we are examining auditory perception & it's testing, not electronic engineering.

 

"So I dont really see why would that kind of triple-sure assurance is needed." What triple a-sure assurance are you talking about? I wouldn't call determining the error rate embedded in DBT results is a triple-sure assurance

Now some of the wilder assertions - DBTs are 99% correct? Isn't that the very point of the discussion? - neither jud nor I accept this assertion without knowing the level of FNs in the results - something that has not been demonstrated. You already have made up your mind on the results as being 99% correct without the necessary evidence & this is why the DBTers are said to operate on a "belief system"

 

Thank you for the summarising of the thought process of those who fear examining DBTs as a suitable test, as usually administered outside of research labs.

 

You stretch and missinterpret my words to push me into some sort of DBT fundamentalists corner. As you wish.

But I repeatedly stated they are just a tool. Only one tool. And I just want them to be used instead of being bashed. Discusing their technical merits and limitations is already much better than the ouright bashing and the my-ears-only atitudes. Why not just keep the thread there instead of this he said, I said that you seem to push lately?

So what about those FNs ?! Any new info?