Much truth in that! Less to fiddle with, as well, which doesn't give the user the ability feel that it was something HE did (other than pay money) to make the music sound the way it does. Gordon Holt made the same prediction about digital audio in general when it was still new to most.
I explained it in my previous posts. About distortions of electric signal VS distortions of sounds.
Yes, a.wayne, (I'm not SY but he quoted my post) I agree with you but I was using the term in the thread title for convenience! Actually, "digital" doesn't exist at all - it's just a convenience & agreed protocol - it all still based on analogue waveforms!
Last edited:
"X-factor" distortions
In reply to 5th element's question, I most certainly am not "against" testing THD or IMD as part of any well thought out test sequence, it's just that the obsessive focus on one aspect of performance while ignoring other important factors can lead to results that are not always good, and sometimes very bad.
For example, in designing crossovers for speakers, it has become standard practice to use higher order filters, because in most cases they help get a more uniform frequency response, better lobing characteristics, reduced comb filtering, and better integration of directionality characteristics along with decreases in distortion.
And yet.... when testing subjects' GSR responses, some of our as of yet un-published research (at Holistic Audio) shows that the higher order filters appear to create gradually rising stress responses in test subjects, while true "time-coherent" response actually seems to reduce stress....even when it goes consciously undetected.
Of special interest: this appears to be the case even when there is a subjective preference for the higher order filter's sonics!
So does this mean it is "bad" to design for uniform frequency response or low THD? Of course not!
It's just that it is very important (for our subliminal system) to design for proper time-coherence even though that factor is generally conceded to be "not noticeable" on a conscious level.
The above was tested using our H3 system, which uses a DEQX as a crossover. It uses FIR filters to create "time accurate" signals, while maintaining steep slopes AND also very linear frequency response. When tested for subliminal stress response using GSR, once the time-coherence issue was resolved, the subjective preference for more linear FR correlated well with the GSR tests, and the more linear the FR was, the better the "subliminal stress response".
From this one might infer that the dominant characteristic from the subliminal point of view is the time-accuracy, with the linearity of FR being a secondary, but still important contributor.
The above information, while speaker-centric, relates to the question about THD then, in that low THD is probably an important issue, but is possibly secondary to other forms of distortion like IMD, or HOM or other as of yet un-named factors, some of which may well go undetected until we start testing the listeners for their subliminal responses.
In any event, to further address 5th Element's points, while THD may not be as important as some think, it remains valid to optimize it, as long as the other distortions (both known and unknown) are optimized at the same time.
To return to the speaker analogies, now that I know time-accuracy is more important, I would first design to get time-accuracy and then focus on getting the linear FR. Fortunately, with the DEQX it is very easy to get the time-response to near perfect AND get the FR to within a fraction of a dB across the whole band.
My thinking is that the same will eventually apply to the amplifier world, where we may be able to find which "x-factors" create more subliminal discomfort, then reduce those. And I would predict it will still be useful to reduce THD... but not as the only indicator to look for.
Now for a digression:
Lest I be taken to task for not mentioning this, it seems crucial to address the directionality issue:
so far it seems like omni-directional systems or broad-band controlled directionality OB designs are the best answers... as long as they are time-accurate.
Our H3 system addresses this from the "virtual point source" (omni) perspective and works exceptionally well.
Now I'm thinking of doing something like a super version of the NaO Note (which is an excellent example of uniformly restricted directionality) but using the DEQX or other FIR filtering to get it highly time-accurate and FR linear with low distortion and good vertical lobing characteristics.
IN SUM, in the same way it's important to optimize FR in speakers, it's important to optimize (reduce) THD in amplifiers, but not at the expense of other factors that may be equally or more important.
Well stated! I most heartily agree.
In reply to 5th element's question, I most certainly am not "against" testing THD or IMD as part of any well thought out test sequence, it's just that the obsessive focus on one aspect of performance while ignoring other important factors can lead to results that are not always good, and sometimes very bad.
For example, in designing crossovers for speakers, it has become standard practice to use higher order filters, because in most cases they help get a more uniform frequency response, better lobing characteristics, reduced comb filtering, and better integration of directionality characteristics along with decreases in distortion.
And yet.... when testing subjects' GSR responses, some of our as of yet un-published research (at Holistic Audio) shows that the higher order filters appear to create gradually rising stress responses in test subjects, while true "time-coherent" response actually seems to reduce stress....even when it goes consciously undetected.
Of special interest: this appears to be the case even when there is a subjective preference for the higher order filter's sonics!
So does this mean it is "bad" to design for uniform frequency response or low THD? Of course not!
It's just that it is very important (for our subliminal system) to design for proper time-coherence even though that factor is generally conceded to be "not noticeable" on a conscious level.
The above was tested using our H3 system, which uses a DEQX as a crossover. It uses FIR filters to create "time accurate" signals, while maintaining steep slopes AND also very linear frequency response. When tested for subliminal stress response using GSR, once the time-coherence issue was resolved, the subjective preference for more linear FR correlated well with the GSR tests, and the more linear the FR was, the better the "subliminal stress response".
From this one might infer that the dominant characteristic from the subliminal point of view is the time-accuracy, with the linearity of FR being a secondary, but still important contributor.
The above information, while speaker-centric, relates to the question about THD then, in that low THD is probably an important issue, but is possibly secondary to other forms of distortion like IMD, or HOM or other as of yet un-named factors, some of which may well go undetected until we start testing the listeners for their subliminal responses.
In any event, to further address 5th Element's points, while THD may not be as important as some think, it remains valid to optimize it, as long as the other distortions (both known and unknown) are optimized at the same time.
To return to the speaker analogies, now that I know time-accuracy is more important, I would first design to get time-accuracy and then focus on getting the linear FR. Fortunately, with the DEQX it is very easy to get the time-response to near perfect AND get the FR to within a fraction of a dB across the whole band.
My thinking is that the same will eventually apply to the amplifier world, where we may be able to find which "x-factors" create more subliminal discomfort, then reduce those. And I would predict it will still be useful to reduce THD... but not as the only indicator to look for.
Now for a digression:
Lest I be taken to task for not mentioning this, it seems crucial to address the directionality issue:
so far it seems like omni-directional systems or broad-band controlled directionality OB designs are the best answers... as long as they are time-accurate.
Our H3 system addresses this from the "virtual point source" (omni) perspective and works exceptionally well.
Now I'm thinking of doing something like a super version of the NaO Note (which is an excellent example of uniformly restricted directionality) but using the DEQX or other FIR filtering to get it highly time-accurate and FR linear with low distortion and good vertical lobing characteristics.
IN SUM, in the same way it's important to optimize FR in speakers, it's important to optimize (reduce) THD in amplifiers, but not at the expense of other factors that may be equally or more important.
Stress can be measured with GSR
An EEG in conjunction with a brain scanner is a more complete way of analyzing for brain coherence and the increase or decreases of Alpha waves (and Theta and Delta). But a simple GSR (Galvanometric Skin Resistance) tester has been show to produce simple test results that are directly correlated with the more complex system mentioned above.
A high incidence of Alpha waves (and Theta and Delta) is a good indicator of a low stress level in the test subject's physiology. The higher the wave amplitude, and the more coherence there is between these waves, the better the physiological and mental performance.
This also correlates with measured changes in skin resistance, as measured with a good GSR system: the healthier and more relaxed the subject, the higher the skin resistance. If stress is induced, the skin resistance starts to drop.
Healthy subjects exhibit high skin resistance, especially when meditating or very relaxed.
Those who are sick tend to show very low skin resistance.
The simplest GSR systems use two fingers connected to the conductive teste elements, and still work pretty well.
GSR is commonly used for bio-feedback as it can be remarkably accurate and is much more comfortable to use than EEG head probes with sticky goo..
A quick primer:
An EEG in conjunction with a brain scanner is a more complete way of analyzing for brain coherence and the increase or decreases of Alpha waves (and Theta and Delta). But a simple GSR (Galvanometric Skin Resistance) tester has been show to produce simple test results that are directly correlated with the more complex system mentioned above.
A high incidence of Alpha waves (and Theta and Delta) is a good indicator of a low stress level in the test subject's physiology. The higher the wave amplitude, and the more coherence there is between these waves, the better the physiological and mental performance.
This also correlates with measured changes in skin resistance, as measured with a good GSR system: the healthier and more relaxed the subject, the higher the skin resistance. If stress is induced, the skin resistance starts to drop.
Healthy subjects exhibit high skin resistance, especially when meditating or very relaxed.
Those who are sick tend to show very low skin resistance.
The simplest GSR systems use two fingers connected to the conductive teste elements, and still work pretty well.
GSR is commonly used for bio-feedback as it can be remarkably accurate and is much more comfortable to use than EEG head probes with sticky goo..
Hi,
You make no sense.
Note I wrote Placebo Effect / Expectation Bias. I do not think any further "careful explanation" is required for any intelligent person without an agenda.
However, I shall indulge you anyway, as you seem to require the instrcution and seem unwilling to acquire it the usual way.
Let us take a Pharmaceutical trial as example.
Two groups are given pills, one group is given the actual medicine, the other known in scientific circles as control group receive what is known in scientific circles as Placebo, basically a Pill that looks like medicine, but has no actual potency (that alone carries interesting challenges, cannot just use sugar-pills, can we now).
The interesting observations are that usually a substantial percentage of the control group who received no medicine report improvements in their condition, even in cases where objective medical analysis disagrees with this, while equally a number of the test group report no improvement even in cases where objective medical analysis suggests such.
So we have cases where sick people subjectively feel better after taking what they believe to be medicine, when in fact they are not better, other actually appearing to get better objectively and still others that get better but do not feel better.
The reasons for these apparent contradictions is simple, expectation bias, or as it is called in the context of pharmaceutical research and general scientific enquiry, the Placebo Effect.
And just because we have moved a test from the medical realm to that of perceived sound it does not disappear and to ignore or deny it is simply unscientific or a mark of underdeveloped facilities in basic logic, if it is not a mark of a agenda.
And it works both ways.
Ciao T
PS, I just notice that what I know as experimenters bias is also called "expectation bias", my use refers to expectations held by the test subject, commonly and inaccurately called placebo effect, which strictly speaking does not exist outside pharmaceutical/medical trials.
You make no sense.
Note I wrote Placebo Effect / Expectation Bias. I do not think any further "careful explanation" is required for any intelligent person without an agenda.
However, I shall indulge you anyway, as you seem to require the instrcution and seem unwilling to acquire it the usual way.
Let us take a Pharmaceutical trial as example.
Two groups are given pills, one group is given the actual medicine, the other known in scientific circles as control group receive what is known in scientific circles as Placebo, basically a Pill that looks like medicine, but has no actual potency (that alone carries interesting challenges, cannot just use sugar-pills, can we now).
The interesting observations are that usually a substantial percentage of the control group who received no medicine report improvements in their condition, even in cases where objective medical analysis disagrees with this, while equally a number of the test group report no improvement even in cases where objective medical analysis suggests such.
So we have cases where sick people subjectively feel better after taking what they believe to be medicine, when in fact they are not better, other actually appearing to get better objectively and still others that get better but do not feel better.
The reasons for these apparent contradictions is simple, expectation bias, or as it is called in the context of pharmaceutical research and general scientific enquiry, the Placebo Effect.
And just because we have moved a test from the medical realm to that of perceived sound it does not disappear and to ignore or deny it is simply unscientific or a mark of underdeveloped facilities in basic logic, if it is not a mark of a agenda.
And it works both ways.
Ciao T
PS, I just notice that what I know as experimenters bias is also called "expectation bias", my use refers to expectations held by the test subject, commonly and inaccurately called placebo effect, which strictly speaking does not exist outside pharmaceutical/medical trials.
Just give everyone a glass of their favourite a few minutes before and again whilst listening, that should help keep things relaxed - I know that helps to work with me!
You seemed to miss the point of my thread.
Thorsten got exactly what I was asking and gave me the type of answer I was after.
As is usual people start nit picking, it was implied that the distortion spectra wasn't objectionable and that both THD and IMD were equally as low as possible.
I started this off by saying that measurements should be used as a way of assessing a designs performance. I am not saying that I would chase low THD just for the sake of low THD in the design phase.
I am saying that if you've got an IC, be it a DAC an ADC etc, it will have a certain level of inherent performance. In this case you should strive to figure out all the bugs of the implementation to arrive at that performance.
If you want to include some analogue stage that has worse overall distortion performance after/before the IC, then you will never realise the performance the IC is capable of. Note that I am not saying this is a bad thing, however, in that situation I would first design a PCB etc with an analogue stage that could realise the performance of the IC. The sole intention of this being to actually come up with an implementation where the IC is working exactly as intended. After having gone this far I would then substitute in the preferred analogue section, be it tubes/passive/discrete etc.
This also applies to an amplifier design. If you've got a blameless three stage as popularised by Self, then yes, with that, in the design phase and then especially in the implementation you should strive to get the most out of the amplifier that you can.
If you want to design a three stage amp of a different variety then that's fine also, it might have worse performance, but this is fine too. Maybe you could add in a couple of extra transistors to improve the measured performance during the design phase, but you don't want to do this either, otherwise it will be a different design and sound worse.
However even if you don't add in those extra transistors, you should still attempt to maximise the overall potential of the design by tweaking component values within the range that you decide as being acceptable.
Then you should attempt to actually realise this performance when it is built.
You seemed to miss the point of my thread.
Thorsten got exactly what I was asking and gave me the type of answer I was after.
As is usual people start nit picking, it was implied that the distortion spectra wasn't objectionable and that both THD and IMD were equally as low as possible.
I started this off by saying that measurements should be used as a way of assessing a designs performance. I am not saying that I would chase low THD just for the sake of low THD in the design phase.
I am saying that if you've got an IC, be it a DAC an ADC etc, it will have a certain level of inherent performance. In this case you should strive to figure out all the bugs of the implementation to arrive at that performance.
If you want to include some analogue stage that has worse overall distortion performance after/before the IC, then you will never realise the performance the IC is capable of. Note that I am not saying this is a bad thing, however, in that situation I would first design a PCB etc with an analogue stage that could realise the performance of the IC. The sole intention of this being to actually come up with an implementation where the IC is working exactly as intended. After having gone this far I would then substitute in the preferred analogue section, be it tubes/passive/discrete etc.
This also applies to an amplifier design. If you've got a blameless three stage as popularised by Self, then yes, with that, in the design phase and then especially in the implementation you should strive to get the most out of the amplifier that you can.
If you want to design a three stage amp of a different variety then that's fine also, it might have worse performance, but this is fine too. Maybe you could add in a couple of extra transistors to improve the measured performance during the design phase, but you don't want to do this either, otherwise it will be a different design and sound worse.
However even if you don't add in those extra transistors, you should still attempt to maximise the overall potential of the design by tweaking component values within the range that you decide as being acceptable.
Then you should attempt to actually realise this performance when it is built.
2 fingers
I've got various systems I'm working with. The 2 finger is reasonably good for "quick and not-so-dirty" work. More accurate is a system using probes that connect to the back of the neck, near the base of the cranium, and to the feet, but that is pretty involved and a little messy, so for much of the work we just stick with the 2 finger variety despite having less overall accuracy.
But 95% vs 70% accuracy isn't really too big an issue when you can clearly and repeatably observe rising or falling trends that closely correlate with the stimulus or lack of such. This is the case, at least with time-accuracy on speakers.
It may be that the higher accuracy (and more expensive!) systems would be needed for testing with amplifiers, where the differences may be much more subtle.
Once you've proven the basic point, and identified a basic issue (like time-accuracy), the rigors associated with doing a test with full statistical analysis seem a little over-blown. I tend to be interested in activities that are sufficiently remunerated, and so, have thus far declined to get all the info together in way that would be deemed worthy of publication.
I would do it if I could get a grant that would pay for the rather considerable amounts of time in doing it with full publishable rigor!
Also, I'm working on a test system that I may try to bring to market, and so would not yet want to reveal too many details about that one or my testing methodology.
But a good 2 finger GSR unit would certainly be a valid and not too expensive way to start validating some of these perspectives... and if nothing else, it's also really great to work with bio-feedback and learn how to consciously relax... God knows I certainly needed that! (And still do...heheh)
Short answer: yes.
I've got various systems I'm working with. The 2 finger is reasonably good for "quick and not-so-dirty" work. More accurate is a system using probes that connect to the back of the neck, near the base of the cranium, and to the feet, but that is pretty involved and a little messy, so for much of the work we just stick with the 2 finger variety despite having less overall accuracy.
But 95% vs 70% accuracy isn't really too big an issue when you can clearly and repeatably observe rising or falling trends that closely correlate with the stimulus or lack of such. This is the case, at least with time-accuracy on speakers.
It may be that the higher accuracy (and more expensive!) systems would be needed for testing with amplifiers, where the differences may be much more subtle.
Once you've proven the basic point, and identified a basic issue (like time-accuracy), the rigors associated with doing a test with full statistical analysis seem a little over-blown. I tend to be interested in activities that are sufficiently remunerated, and so, have thus far declined to get all the info together in way that would be deemed worthy of publication.
I would do it if I could get a grant that would pay for the rather considerable amounts of time in doing it with full publishable rigor!
Also, I'm working on a test system that I may try to bring to market, and so would not yet want to reveal too many details about that one or my testing methodology.
But a good 2 finger GSR unit would certainly be a valid and not too expensive way to start validating some of these perspectives... and if nothing else, it's also really great to work with bio-feedback and learn how to consciously relax... God knows I certainly needed that! (And still do...heheh)
I'll continue reading momentarily, but had to pause and point out "So we have cases where sick people subjectively feel better after taking what they believe to be medicine, when in fact they are not better, other actually appearing to get better objectively and still others that get better but do not feel better."
If oncologists perform this test, then those people who got better but still feel ill are no longer the oncologists patients. They're a psychiatrist's patients.
In other words, if the DBT is concerned with whether pill 319 reduces the size of a tumor, how the person feels is of no concern.
If you want a DBT concerned with how a person feels affecting tumor size, that's all well and good also. But let's not confuse the two.
If oncologists perform this test, then those people who got better but still feel ill are no longer the oncologists patients. They're a psychiatrist's patients.
In other words, if the DBT is concerned with whether pill 319 reduces the size of a tumor, how the person feels is of no concern.
If you want a DBT concerned with how a person feels affecting tumor size, that's all well and good also. But let's not confuse the two.
If I understand correctly, measurement junkies are being denounced with... more measurements! What has me most curious is, what if my alpha pleasure waves show I love it, but my conscious hearing says I hate it?
I think I need one hi-fi system for big hall orchestra stuff, another for small combo electric rock stuff, still another for solo acoustic stuff, etc.
I think I need one hi-fi system for big hall orchestra stuff, another for small combo electric rock stuff, still another for solo acoustic stuff, etc.
I want to stress few points.
High order filters are well audible even on conscious level. They sound exactly like ringing sounds.
Second, when I say about subconscious reactions, I mean reactions on sounds that correspond to material reproduced, like water runs suddenly in the room causing me to jump and turn around before I recognize that it is not in my room, it is on CD.
And third, statistics can't be used instead of knowledge of laws of nature. As many experiments particularly carried by Institute of Noetic Sciences demonstrate, generators of random numbers installed in computers in universities worldwide are somehow linked with human thoughts. This fact is real, despite of absence of scientific explanations.
They discovered this phenomenon a first doing experiments you do, to test subconscious reactions on stimuli randomly selected by computers. A first they tried to explain this by errors of measurements, but going further and further they made experiments genuinely cost effective: data produced by random generators is recorded in database and when some strange changes happen they search for significant worldwide events that drew attention of people causing some kind of mental and emotional synchronization. It correlates well, that is very strange and unexplainable.
High order filters are well audible even on conscious level. They sound exactly like ringing sounds.
Second, when I say about subconscious reactions, I mean reactions on sounds that correspond to material reproduced, like water runs suddenly in the room causing me to jump and turn around before I recognize that it is not in my room, it is on CD.
And third, statistics can't be used instead of knowledge of laws of nature. As many experiments particularly carried by Institute of Noetic Sciences demonstrate, generators of random numbers installed in computers in universities worldwide are somehow linked with human thoughts. This fact is real, despite of absence of scientific explanations.
They discovered this phenomenon a first doing experiments you do, to test subconscious reactions on stimuli randomly selected by computers. A first they tried to explain this by errors of measurements, but going further and further they made experiments genuinely cost effective: data produced by random generators is recorded in database and when some strange changes happen they search for significant worldwide events that drew attention of people causing some kind of mental and emotional synchronization. It correlates well, that is very strange and unexplainable.
No. It is plainly wrong. If it was true we all should be treated by psychiatrists only, for any illness, including injures. It is well known fact that nurses in hospitals during wars were selected by psychological influence on speed of healing of injured soldiers. When so called traditional medicine is used together with modern scientific-based medicine, results are better, even in oncology. Such research were conducted again and again, and best doctors are well aware about that.
PS: my wife, by the way, participates in many trails related to diabetes as a coordinator.
I would say, they are very strange devices. The inventor of polygraph was considered being crazy because he tried to prove that sensors connected to plant leaves in his office register his emotions on any distance.
No, and you've provided a non sequitor as proof. If the cancer no longer remains but the patient feels no better, there is still some non-oncological illness. Perhaps hypochondria. I've made no claim that patient state of mind has no effect on health. And it may well have an effect on hi-fi listening also.
Last edited:
- Status
- Not open for further replies.