Serious Question: How can DAC's have a SOUND SIGNATURE if they measure as transparent? Are that many confused?

[Deleted member 12179]

You can measure those. You can measure brain wave behavior. You can measure what chemical released in your body while enjoying. You can measure all aspects of sound. If you know how to interpret such measurements you still have to implement such thing. It's difficult yes. But it's possible. Measurement is not something that's limiting all of this.

Oh, that seems great then, but I have never seen that kind of data ever in audio reviews, no? I´ve seen those brain reading helmets kinda thing. Those kind of measurements would be really interesting. And if in all those brain waves, if it would be easy to completely separate and categorize what pleasure comes from the music, melodies, sound complexity, which one from the pure audio quality transmission and detail. ABX pleasure measurements. "Objective audio-pleasure neuro-measurements", works as album title too.

They say in the future dreams will be easily recorded and backed up. In the future, but now it's f. response, off axis representation. All really important for choosing gear.

 

[JohnYang1997]

Oh, that seems great then, but I have never seen that kind of data ever in audio reviews, no? I´ve seen those brain reading helmets kinda thing. Those kind of measurements would be really interesting. And if in all those brain waves, if it would be easy to completely separate and categorize what pleasure comes from the music, melodies, sound complexity, which one from the pure audio quality transmission and detail. ABX pleasure measurements. "Objective audio-pleasure neuro-measurements", works as album title too.

They say in the future dreams will be easily recorded and backed up. In the future, but now it's f. response, off axis representation. All really important for choosing gear.

Those measurements are only to explain why ONE person like something or not.
If you want to simply know what ONE like, you can ask him. Hence all the controlled listening tests.

 

[Kal Rubinson]

Agreed, but it might be wise to spend a day or two listening to both A & B to familiarize yourself before doing an A/B/X

Certainly.

 

[andreasmaaan]

There is an objective component to double blind comparisons and, to a lesser extent, even to direct transparent comparison but it's not the same and a well designed and objective testing regime.

For example, the flattest speakers tend to not bring the most pleasing sound.

The largest body of research that's been done on this topic pretty clearly contradicts that statement, though.

 

[StevenEleven]

Oh, that seems great then, but I have never seen that kind of data ever in audio reviews, no? I´ve seen those brain reading helmets kinda thing. Those kind of measurements would be really interesting. And if in all those brain waves, if it would be easy to completely separate and categorize what pleasure comes from the music, melodies, sound complexity, which one from the pure audio quality transmission and detail. ABX pleasure measurements. "Objective audio-pleasure neuro-measurements", works as album title too.

Here's a start!

https://www.nih.gov/news-events/news-releases/our-brains-appear-uniquely-tuned-musical-pitch

 

[raistlin65]

The largest body of research that's been done on this topic pretty clearly contradicts that statement, though.

The reliability of that research is now suspect. It is all built on that 2004 study where Sean Olive tested listeners with a large number of loudspeakers and concluded that their preference was generally neutral.

However, in 2015, after writing about the research where listeners adjusted headphone sound using bass and treble controls, he admitted that the non-neutral speakers in that original study had other flaws which affected the results. And he admits that people have some preference for treble and bass adjustment over neutral with speakers, which makes sense based upon all of their findings regarding the Harman Curve.

Here are the comments he posted to his blog about this
"The early studies involved comparison of different speakers that varied more than bass and treble balance. Some speakers had resonances that produced serious colorations, distortions, differences in directivity. The headphone study basically takes a flat neutral headphone and asks people to adjust the bass and treble. That's where experience and age seem to take over. The same holds true for loudspeakers when we did a similar experience."

"Not really. Prior to this study, I nor anyone I know had published a study where trained and untrained listeners were given a bass and treble control and asked to adjust to taste. In previous studies, trained and untrained listeners were asked to give preference ratings to speakers that varied in ways other than bass and treble. It seems that given some finite choices people will pick the most neutral speaker or headphone (no resonances), wide bandwidth. However, given some tone controls they will adjust for variations in program and taste.

This study does confirm a previous one where we found untrained listeners adjusted the bass and treble of a loudspeaker and headphone higher than trained listeners. That study only had 3 untrained listeners, which is why we did this much larger study.

http://www.aes.org/e-lib/browse.cfm?elib=17042"

Edit: Updated to provide the link to the comments http://seanolive.blogspot.com/2015/11/factors-that-influence-listeners.html

 

[Julf]

I WANT TO BELIEVE !!!!!!

I thing churches exist specifically for that purpose...

 

[andreasmaaan]

The reliability of that research is now suspect. It is all built on that 2004 study where Sean Olive tested listeners with a large number of loudspeakers and concluded that their preference was generally neutral.

However, in 2015, after writing about the research where listeners adjusted headphone sound using bass and treble controls, he admitted that the non-neutral speakers in that original study had other flaws which affected the results. And he admits that people have some preference for treble and bass adjustment over neutral with speakers, which makes sense based upon all of their findings regarding the Harman Curve.

Here are the comments he posted to his blog about this
"The early studies involved comparison of different speakers that varied more than bass and treble balance. Some speakers had resonances that produced serious colorations, distortions, differences in directivity. The headphone study basically takes a flat neutral headphone and asks people to adjust the bass and treble. That's where experience and age seem to take over. The same holds true for loudspeakers when we did a similar experience."

"Not really. Prior to this study, I nor anyone I know had published a study where trained and untrained listeners were given a bass and treble control and asked to adjust to taste. In previous studies, trained and untrained listeners were asked to give preference ratings to speakers that varied in ways other than bass and treble. It seems that given some finite choices people will pick the most neutral speaker or headphone (no resonances), wide bandwidth. However, given some tone controls they will adjust for variations in program and taste.

This study does confirm a previous one where we found untrained listeners adjusted the bass and treble of a loudspeaker and headphone higher than trained listeners. That study only had 3 untrained listeners, which is why we did this much larger study."

http://www.aes.org/e-lib/browse.cfm?elib=17042"

From the study you linked:

"Above 100-300 Hz, the shape of the preferred in-room loudspeaker closely matches the predicted in-room response of the loudspeaker (see Fig. 1). This means that as long as the loudspeaker is well designed (i.e. having a flat on-axis response, and smooth off-axis response and directivity), it shouldn’t require equalization above 100-300 Hz to meet the preferred inroom target response curve."

I agree with you 100% if we confine the discussion to the bass, where the room and not the speaker is the dominant factor anyway.

But in the frequency range where the speaker is dominant, this study agrees with his prior finding that listeners tend to prefer flat on-axis response and smooth off-axis response and directivity, i.e. a neutral speaker. Of course, individual preferences vary, but it is upon neutrality that they (at least according to Olive) tend to converge.

Regarding your reference to the Harman curve, are you referring to the headphone target curve here?

 

[TomB19]

However, in 2015, after writing about the research where listeners adjusted headphone sound using bass and treble controls, he admitted that the non-neutral speakers in that original study had other flaws which affected the results.

Good point.

His "non-linear" speakers are speakers with fairly sharp peaks and holes in the response curve. In no way does it compare a perfectly flat response to the BBC curve or some other previously discovered response contour. He is just saying a better speaker is preferred to an inferior one.

Next, the study appears to completely ignore phase coherence of the speakers under test. Given the response curves, particularly of B and M, surely there are phase issues.

Also, I may have missed the foundational component of the research but I'm skeptical it is a "larger body" of research than generated by the BBC or other industry research which it contests.

 

[andreasmaaan]

Next, the study appears to completely ignore phase coherence of the speakers under test. Given the response curves, particularly of B and M, surely there are phase issues.

Other studies have failed to find this is audible in the magnitude in which it is present in conventional speakers, except insofar as it effects amplitude response.

In isolation from that, it's either negligible or completely irrelevant.

 

[raistlin65]

From the study you linked:

"Above 100-300 Hz, the shape of the preferred in-room loudspeaker closely matches the predicted in-room response of the loudspeaker (see Fig. 1). This means that as long as the loudspeaker is well designed (i.e. having a flat on-axis response, and smooth off-axis response and directivity), it shouldn’t require equalization above 100-300 Hz to meet the preferred inroom target response curve."

I agree with you 100% if we confine the discussion to the bass, where the room and not the speaker is the dominant factor anyway.

But in the frequency range where the speaker is dominant, this study agrees with his prior finding that listeners tend to prefer flat on-axis response and smooth off-axis response and directivity, i.e. a neutral speaker. Of course, individual preferences vary, but it is upon neutrality that they (at least according to Olive) tend to converge.

Regarding your reference to the Harman curve, are you referring to the headphone target curve here?

My bad. I forgot to provide the link to his comments above (edited to add them): http://seanolive.blogspot.com/2015/11/factors-that-influence-listeners.html

So you have to read his comments in the context that in 2015, he's walking back that users prefer a strictly linear response, that previous research was flawed due to the sample set of speakers.

 

[raistlin65]

Good point.

His "non-linear" speakers are speakers with fairly sharp peaks and holes in the response curve. In no way does it compare a perfectly flat response to the BBC curve or some other previously discovered response contour. He is just saying a better speaker is preferred to an inferior one.

Next, the study appears to completely ignore phase coherence of the speakers under test. Given the response curves, particularly of B and M, surely there are phase issues.

Also, I may have missed the foundational component of the research but I'm skeptical it is a "larger body" of research than generated by the BBC or other industry research which it contests.

Exactly. They used a poor sample set. A reader on his blog specifically asks if the headphone research of 2015 about users adding tone controls contradicts the results of the earlier loudspeaker research, and he says that "The early studies involved comparison of different speakers that varied more than bass and treble balance. Some speakers had resonances that produced serious colorations, distortions, differences in directivity." http://seanolive.blogspot.com/2015/11/factors-that-influence-listeners.html

Sean Olive and Harman aren't going to publish a press release that "Hey our earlier research was flawed."

 

[andreasmaaan]

Sorry. I put the wrong link. Here is the link to his comments on his blog: http://seanolive.blogspot.com/2015/11/factors-that-influence-listeners.html

So 2015, after his loudspeaker research that everyone cites, he's admitting there are problems with the sample set in speaker selection.


My bad. I forgot to provide the link to his comments above (edited to add them): http://seanolive.blogspot.com/2015/11/factors-that-influence-listeners.html

So you have to read his comments in the context that in 2015, he's walking back that users prefer a strictly linear response, that previous research was flawed due to the sample set of speakers.

Thanks, that's clearer now.

I am familiar with this more recent Olive study, but I don't believe it's findings contradict his previous work at all in respect of loudspeakers.

What happened in the loudspeaker portion of this larger 2013 study (let's put the headphones aside for the purposes of this discussion) was that listeners were asked to adjust bass and trebles controls to preference from the starting point of the flattened steady-state in-room response of a neutral loudspeaker.

Here is the predicted unflattened in-room steady-state response of the loudspeaker used (blue trace) from one of the previous studies, along with the average preferred in-room response as determined in that same study (black trace). You'll note that the two are virtually identical from 250-20,000Hz:

And here is the FR of the same speaker after EQ has been applied to flatten the in-room response (green dashed trace):

From the above, it must be inferred that, in order to flatten the speaker's steady-state response, the treble was shelved up by around 2-3dB from around 1kHz, and the bass was shelved down by about 2-3dB below around 200Hz.

This flattened in-room response, with its shelved-up treble / shelved-down bass, was what was used as the reference in the 2013 study. Not the response of an unequalised neutral speaker in a room, but rather the response of a speaker with shelved up treble and shelved-down bass.

This a good point to display the shelving filters used in the 2013 study. You'll notice that in terms of Q they correspond to the in-room response curve preferred by listeners in that previous study:

Looking at those shelving filters, and looking at the preferred response from the earlier study, you would predict loudspeaker listeners in the 2015 study to shelve the base up by about 6dB and the treble down by about 3dB in order to reach the in-room frequency response of a neutral loudspeaker with an additional bass boost.

And indeed, this is what listeners, on average, did in 2013:

In other words, if you took a system with a flat in-room steady-state response and applied the bass and treble shelves that listeners applied on average in the 2013 study, you'd end up with something very similar to the preferred curve (black trace, above) from the earlier studies, which in turn tends to converge that of a neutral speaker placed in a room (except in the range below about 250Hz).

 

[andreasmaaan]

Exactly. They used a poor sample set. A reader on his blog specifically asks if the headphone research of 2015 about users adding tone controls contradicts the results of the earlier loudspeaker research, and he says that "The early studies involved comparison of different speakers that varied more than bass and treble balance. Some speakers had resonances that produced serious colorations, distortions, differences in directivity." http://seanolive.blogspot.com/2015/11/factors-that-influence-listeners.html

Sean Olive and Harman aren't going to publish a press release that "Hey our earlier research was flawed."

This is, but this doesn't disprove the research findings of the 2013 study, which remains the most comprehensive study of its kind. If further research comes along in future that tends to disprove the findings, we should have open minds. But for now, this is by far the best that we have.

Plus, see my previous post. The 2013 study tends to show that average preferences converged with those of the earlier studies.

 

[raistlin65]

This is, but this doesn't disprove the research findings of the 2013 study, which remains the most comprehensive study of its kind. If further research comes along in future that tends to disprove the findings, we should have open minds. But for now, this is by far the best that we have.

Plus, see my previous post. The 2013 study tends to show that average preferences converged with those of the earlier studies.

What is the title and/or link to 2013 loudspeaker study? I only know of the headphone study from 2013: http://seanolive.blogspot.com/2013/04/

 

[andreasmaaan]

What is the title and/or link to 2013 loudspeaker study? I only know of the headphone study from 2013: http://seanolive.blogspot.com/2013/04/

This is the one. Listeners were asked to adjust bass and treble filters on a Revel F208 that had been EQ'd initially to have a flat in-room response.

On average, listeners EQ'd the treble filter to the point where it was very close to anechoically flat (resulting in a -2.4dB treble roll-off in the steady-state response), and EQ'd the bass to the point where it was shelved up by about 6dB below 200-ish Hz.

This resulted in an average preferred EQ curve that very tightly matched the predicted in-room response of the speaker from 250Hz upwards:

Of course, I agree that this study leaves plenty of unanswered questions, but it does seem that at the very least that it failed to falsify the findings of the previous work.

 

[raistlin65]

This is the one. Listeners were asked to adjust bass and treble filters on a Revel F208 that had been EQ'd initially to have a flat in-room response.

On average, listeners EQ'd the treble filter to the point where it was very close to anechoically flat (resulting in a -2.4dB treble roll-off in the steady-state response), and EQ'd the bass to the point where it was shelved up by about 6dB below 200-ish Hz.

This resulted in an average preferred EQ curve that very tightly matched the predicted in-room response of the speaker from 250Hz upwards:

View attachment 50870

Of course, I agree that this study leaves plenty of unanswered questions, but it does seem that at the very least that it failed to falsify the findings of the previous work.

I can't access the article (not paying $33). But this makes more sense as it shows that people NOT tend to prefer a flat in-room response. Moreover, I see from the abstract "There were significant variations in the preferred bass and treble levels." Those variations seem important, along with the methodology for selecting the listeners (which I cannot access). And, what kind of EQ options the users were given.

And TomB19 was not wrong when he stated the following, which you told him was incorrect

For example, the flattest speakers tend to not bring the most pleasing sound.

 

[andreasmaaan]

I can't access the article (not paying $33). But this makes more sense as it shows that people NOT tend to prefer a flat in-room response. Moreover, I see from the abstract "There were significant variations in the preferred bass and treble levels." Those variations seem important, along with the methodology for selecting the listeners (which I cannot access). And, what kind of EQ options the users were given.

I posted a picture of the EQ curves in post #773. All the pics in that post are taken from that article, btw.

The listeners in the study were mostly people with professional experience in audio, some "trained" by Harman and other not. This is obviously a major shortcoming of the study.

And TomB19 was not wrong when he stated the following, which you told him was incorrect

I didn't say that the statement was incorrect, I said that the most comprehensive evidence we have to date does not support the statement. Just to be clear on that point.

I then fleshed out my argument by explaining that the evidence that we do have suggests that, on average, listeners prefer a speaker that is neutral above 300-ish Hz (in retrospect, I maybe should have said 250Hz).

I then agreed with his statement to the extent that the evidence suggests that, on average, listeners preferred a speaker to have a slightly exaggerated bass response.

I noted, however, that the room dominates the response in this region, not the speaker.

Throughout, I noted that the evidence we have at present is far from conclusive and that I'm keeping an open mind for fresh evidence. Perhaps you have some to offer?

 

[raistlin65]

I didn't say that the statement was incorrect, I said that the most comprehensive evidence we have to date does not support the statement. Just to be clear on that point.

Sure his statement did. He said "the flattest speakers" and the 2013 study sounds like it supports what he says.

Throughout, I noted that the evidence we have at present is far from conclusive and that I'm keeping an open mind for fresh evidence. Perhaps you have some to offer?

I am a former academic, and I am skeptical of corporate-based research, particularly research, like Olive's, that tends not to share important information. Such as when they did the large study of 40 loudspeakers but didn't share which speakers they were. Now we find out that the non-neutral speakers had other issues. Corporate based research does not have the same ethical and rigor standards as academic research (largely) does.

Meanwhile, the standard deviation of that 2013 study is just as important as the sample group. And you have to ask yourself, if Olive is such a good researcher, why wouldn't he get a good sample group for that 2013 study? Harman can afford to pay participants. I think the researcher has a problem with good methodology, although perhaps he's realized that due to the 2015 headphones study which did have a much better set of listeners.

So I'm not on the "take the sound byte version of Sean Olive's research and run with it as if it applies to everyone" track, as many seem want to do.

 

[andreasmaaan]

Sure his statement did. He said "the flattest speakers" and the 2013 study sounds like it supports what he says.

Could you elaborate on this? The 2013 study found that, when presented with a speaker that had been equalised to not be flat, listeners EQ'd it themselves so that it was flat (from 250Hz upwards).

Somehow we seem to be misunderstanding each other on this point. I can't work out where the misunderstanding lies.

I am a former academic, and I am skeptical of corporate-based research, particularly research, like Olive's, that tends not to share important information. Such as when they did the large study of 40 loudspeakers but didn't share which speakers they were. Now we find out that the non-neutral speakers had other issues.

I agree with your general criticism re: corporate-funded research.

Nevertheless, in this case, I think your specific criticisms are misguided.

Firstly, I would not expect a player in the market to share which speakers were studied, as it would potentially open them up to the risk of litigation.

Secondly, the original studies were clear on the fact that the non-neutral speakers in the study had other issues, as you put it. This is not some hidden piece of information that was only revealed post facto.

But, like I said earlier, these studies are not perfect, and as you say, they were not undertaken by the perfect group of researchers (being interested parties). There is definitely more to learn. But until that happens, I don't see a lot of solid evidence that the findings are false. Could you point me to any?