The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Launch of Pono Studio Headphones
Old 11th April 2014
  #2821
Gear Addict
Quote:
Originally Posted by chrisj View Post
Apple does the luxury thing of well designed costly gear, including portable music players. Pono takes the luxury segment of Apple's segment of that market, which doesn't really damage Apple but is more than enough for a company Pono's size to thrive.
I'm afraid I don't share your optimism with regard to Pono as a going concern, if the Apple rumors prove to be correct. iTunes and Apple music players are already so entrenched in the market, Pono is just vaporware at this point, and their prototype player has an inconvenient form factor and no apparent advantages over a next-gen iPhone or iPod with hi-res capability. Not to mention the other portable hi-res players already available from FiiO, Astell & Kern, and others.

Sorry, but at this point I'd be VERY surprised if Pono even gets off the ground. A new company, no matter how well-funded (and really, $8 million isn't that much in today's world), can't begin to approach the marketing reach and power of Apple. Stick a fork in it. It's done.

Cheers,
Eddie
Old 11th April 2014
  #2822
Airwindows
 
chrisj's Avatar
Quote:
Originally Posted by theblue1 View Post
Let me be clear: I did not mean to suggest that you were trying to say that I hadn't written what I'd written -- but rather it seemed to me that you were suggesting that your quote of it -- with those notable omissions -- was somehow still equivalent to what I'd written and could serve as a new baseline.

(And you will note that that 'rephrasing' took on a life of its own as others shot it down. At the point where I felt I had to explain my original post to Alistair, my patience was wearing thin with the situation.)
Yeah, that's true. Me, I'd be down with the rephrase as a new baseline, still. I wasn't trolling you, I do mean that and I think it's the case. But I do also remember seeing people associating it with you—and it was wrong of me to allow that situation to happen, when you were approaching things with great seriousness.

I am sorry.

I'm also just as serious, even though I can usually avoid getting upset by the stuff people say about me and the points I'm making. But I really do think we're looking at a cool audio reinassance that might help us ALL out, where the opposing camp seems mostly mad that 'people are being sold snake oil' and want to restrict people so they can only get what they 'should' need.

I can be passionately and intensely against that point of view without being too playful, and I will play only with Kenny from now on. Truce?
Old 11th April 2014
  #2823
Quote:
Originally Posted by chrisj View Post
Yeah, that's true. Me, I'd be down with the rephrase as a new baseline, still. I wasn't trolling you, I do mean that and I think it's the case. But I do also remember seeing people associating it with you—and it was wrong of me to allow that situation to happen, when you were approaching things with great seriousness.

I am sorry.

I'm also just as serious, even though I can usually avoid getting upset by the stuff people say about me and the points I'm making. But I really do think we're looking at a cool audio reinassance that might help us ALL out, where the opposing camp seems mostly mad that 'people are being sold snake oil' and want to restrict people so they can only get what they 'should' need.

I can be passionately and intensely against that point of view without being too playful, and I will play only with Kenny from now on. Truce?
Truce? Oh, absolutely. I was planning on PMing you to try to repair fences. It just takes me a while to uncoil from striking position. heh

I think my superficial charm can sometimes mask a certain sort of seething anger that sometimes builds in me; I'm determined to learn to deal with if it takes 'til I'm 90. (Won't be that long, anyhow.)

We're good on this.
Old 11th April 2014
  #2824
Gear Guru
 

Quote:
Originally Posted by Eddie TX View Post
I'm afraid I don't share your optimism with regard to Pono as a going concern, if the Apple rumors prove to be correct. iTunes and Apple music players are already so entrenched in the market, Pono is just vaporware at this point, and their prototype player has an inconvenient form factor and no apparent advantages over a next-gen iPhone or iPod with hi-res capability. Not to mention the other portable hi-res players already available from FiiO, Astell & Kern, and others.

Sorry, but at this point I'd be VERY surprised if Pono even gets off the ground. A new company, no matter how well-funded (and really, $8 million isn't that much in today's world), can't begin to approach the marketing reach and power of Apple. Stick a fork in it. It's done.

I can't find it now, but I believe there was something in the literature from some of the Pono people that was saying high-resolution was their "cause" and that if they got hi-res music "off the ground" or onto people's radar, that they would be happy.

Perhaps they will see Apple's entry into this field as the de facto achievement of their goals and declare "victory" and never actually build or sell the Pono device itself. Certainly that statement I saw gave me the impression that would be OK with them - or some of them.

Quote:
Not to mention the other portable hi-res players already available from FiiO, Astell & Kern, and others.
Yes what kills me is the comments section where people are saying they "can't wait" until October to get their Pono! Well if they can't wait, why are they waiting? Same with the people in this thread rhapsodizing about how Pono WILL CHANGE the face of 'how people listen to music', when such products already exist (FiiO, et al) and the content for them already exists (HD Tracks etc) and apparently the demand for them already exists (the comments section)

Why is it not changing already?

Will Apple "ruin" the grooviness of Pono? Is Pono better because it is a small company, or better because Neil Young is involved, or better because its player is an inconvenient shape that will force you to sit down and listen to music? If it is TRULY 192/24 that will save the world, what's wrong with 192/24 from the Goliath Apple vs 192/24 from the David Pono? You live by the sample rate, you die by the sample rate.
Old 11th April 2014
  #2825
Lives for gear
 
O.F.F.'s Avatar
 

Quote:
Originally Posted by joeq View Post
I can't find it now, but I believe there was something in the literature from some of the Pono people that was saying high-resolution was their "cause" and that if they got hi-res music "off the ground" or onto people's radar, that they would be happy.

Perhaps they will see Apple's entry into this field as the de facto achievement of their goals and declare "victory" and never actually build or sell the Pono device itself. Certainly that statement I saw gave me the impression that would be OK with them - or some of them.

I seem to remember that it was Neil Young himself who said that.
Old 11th April 2014
  #2826
Gear Guru
 
Karloff70's Avatar
 

Quote:
Originally Posted by joeq View Post
Why is it not changing already?
Because it hadn't gathered any momentum with the masses? Looks like it will now, certainly if Apple are doing it. Got issue with that? Is that bad? Is it hard to understand that the Pono guys just want better audio to become the standard? Makes complete sense from here somehow. Did from the beginning.

Sometimes it beggers belief how arguing for the sake of it about 'what's wrong' makes everything else disappear around here.
Old 11th April 2014
  #2827
Old 11th April 2014
  #2828
Lives for gear
 
paul brown's Avatar
Quote:
Originally Posted by CarmenC View Post
95 pages of discussion about an audio player with no specs available. their marketing department has done very well to deflect the need for solid evidence!
Old 11th April 2014
  #2829
Lives for gear
 
3rd Degree's Avatar
 

Quote:
Originally Posted by paul brown View Post
95 pages of discussion about an audio player with no specs available. their marketing department has done very well to deflect the need for solid evidence!
Evidence? Since when did companies release specs of an unreleased product that sounds to be in the prototype stage?
Old 11th April 2014
  #2830
Lives for gear
 
nuthinupmysleeve's Avatar
 

OMG.. this is awesome! It's pretty amazing how applicable to this thread this is.

Old 11th April 2014
  #2831
Quote:
This design called for a small, portable structure capable of storing high-resolution digital files and converting them to analog music, thereby providing a more-than-fulfilling experience for the listener.
Why didn't anyone think of this before??!?

heh heh heh

With regards to specs, I get it, they're still heurizing their way to as near perfection as they can get in the next month before sending whatever they have then to manufacturing. Judging from the specs of other modern, well designed playback gear, I think it's probably fairly safe to say that there shouldn't be any unpleasant surprises for those with realistic expectations.


With regard to the seven perpendicular red line video... seen it before, hilarious. I knew there was something about the corporate world i missed: meetings where people pretend they know what's going on but, of course, are profoundly clueless, are probably already in way over their heads in their own jobs but go on to tell the experts how their field of expertise 'should' work. Not sure how that's pertinent here exactly... heh
Old 11th April 2014
  #2832
Beatport has always offered both mp3 version and WAV for every purchase. You pay more for WAV. Professional DJ's, and not all of them, buy the WAV's because they feel mp3's don't sound as good on large club sound systems. I doubt many ordinary customers buy the WAV option.
The same for iTunes. I doubt many customers will choose the 96khz option unless persuaded by some compelling reason to do so.
I don't think anything much changes for the music industry, which by the way saw a dip in digital sales of 13% this first quarter of 2014.
Old 11th April 2014
  #2833
Gear Guru
 
UnderTow's Avatar
Quote:
Originally Posted by theblue1 View Post
(And you will note that that 'rephrasing' took on a life of its own as others shot it down. At the point where I felt I had to explain my original post to Alistair, my patience was wearing thin with the situation.)
To be clear, I always understood what you wrote and was fully aware that chrisj had misunderstood you and removed the essential qualifiers to come up with his new "baseline" that did not represent what you or anyone else meant.

I treated the quoted bit fully as chrisj's concept even though he used your words to put if forward.

Quote:
Originally Posted by chrisj View Post
Me, I'd be down with the rephrase as a new baseline, still
As long as you present it as your own concept and accept that removing the qualifiers changes the meaning of what theblue1 wrote...


Alistair
Old 11th April 2014
  #2834
Gear Addict
Quote:
Originally Posted by theblue1 View Post
With regard to the seven perpendicular red line video... seen it before, hilarious.
I thought that was a documentary. I guess they didn't have time for the part where, after the meeting, the participants go to the engineer's boss and complain that he's not doing his job properly, then the engineer gets called on the carpet, reprimanded, written up, demoted, etc. You think I'm making this up? No. Happens all the time.

Back to Pono. If Apple kills it before it takes off, I wonder what will happen to the backers' investments? Will they get their promised player, or a refund? Should be interesting to see how the end game plays out.

Cheers,
Eddie
Old 11th April 2014
  #2835
Quote:
Originally Posted by UnderTow View Post
To be clear, I always understood what you wrote and was fully aware that chrisj had misunderstood you and removed the essential qualifiers to come up with his new "baseline" that did not represent what you or anyone else meant.

I treated the quoted bit fully as chrisj's concept even though he used your words to put if forward.



As long as you present it as your own concept and accept that removing the qualifiers changes the meaning of what theblue1 wrote...


Alistair
FWIW, I think we've got it all worked out. I'm happy to take the majority of the weight for getting overwound on the issue.
Old 11th April 2014
  #2836
Quote:
Originally Posted by Eddie TX View Post
I thought that was a documentary. I guess they didn't have time for the part where, after the meeting, the participants go to the engineer's boss and complain that he's not doing his job properly, then the engineer gets called on the carpet, reprimanded, written up, demoted, etc. You think I'm making this up? No. Happens all the time.

Back to Pono. If Apple kills it before it takes off, I wonder what will happen to the backers' investments? Will they get their promised player, or a refund? Should be interesting to see how the end game plays out.

Cheers,
Eddie
I'm pretty sure the player product will come out. I mean, what's to stop it? It may represent a refinement on prior designs and maybe a mid-priced product that can compete with existing high end players, but it's hardly a groundbreaking product.

Now, with regard to the store, arrangements with labels, not sure how that would be affected by the 800 pound gorilla coming into the HD media market. That might be a little less certain. If I had my choice of trying to bring out an HD media player and trying to create a viable HD media store and cut new deals with labels to extend the selection (and hopefully verify the provenance) of HD media, I would MUCH rather tackle the player. But then I learned long ago I'd rather deal with technology than music biz suits any day.
Old 11th April 2014
  #2837
Gear Guru
 

Quote:
Originally Posted by Karloff70 View Post
Because it hadn't gathered any momentum with the masses? Looks like it will now, certainly if Apple are doing it. Got issue with that? Is that bad?
No the only 'issue' I have is noticing that some pro-Pono people seem to be objecting to Apple's entry into this area. If it is true that Neil Young himself would be happy to see that happen, that sort of indicates that what is at work with those people is not a sincere desire for better audio, but something more like a desire for exclusivity and audio snobbery.

These people who supposedly want Better Sound so bad, have heretofore ignored the existing options for those resolutions, and only became interested when some rock stars endorsed an oddly shaped player that looks like nothing else.... so that everyone will know what it is you are toting; and What That Says About You as a music lover. Then they get mad if 'everybody' can have the same quality right on their damn phone!

Quote:
Is it hard to understand that the Pono guys just want better audio to become the standard?.
Not hard to understand. God bless them if they are willing to go to all the trouble of starting a business just to force others into that market. I was just commenting on EddieTX's post that Apple is likely to "bury" Pono as a competitive product and the reactions to that news around the net. Better audio is a good cause.

But the objections from the recording engineers here is that they are bull****ting about the relationship of sample rate to better sound, bull****ting about where the biggest differences are found, and bull****ting is bad even in a "good cause". Like a self-styled revolutionary blowing people up in the name of The People.

We all know what is needed for truly better sound, and Octuple Sample Rates is not it.
Old 11th April 2014
  #2838
If you dare to think rationally about all this, you can't really begin to imagine anyone beyond the "self-proclaimed audiophile hobbyist" crowd even beginning to think about slapping down hundreds of dollars'es for this single-purpose device, an audio player machine.

Because it sounds better. You don't say.

I would bet the only conceivable realistic future for it is if they tack with the breeze and create the Pono app for your tablet or iPad or iPhone or integrate the whole concept into the world as it is.

As an expensive, proprietary unit that solves a problem no one is complaining about, aside from a severely disgruntled hobbyist niche as already referenced, sorry. But-- don't let it bring you down.
Old 11th April 2014
  #2839
Gear Guru
 

Quote:
Originally Posted by joelpatterson View Post
If you dare to think rationally about all this...
see, that's your problem right there
Old 11th April 2014
  #2840
Quote:
Originally Posted by joeq View Post
see, that's your problem right there
one of many!!!
Old 11th April 2014
  #2841
Pono app.


JP may have offered that as an absurdity -- but anyone who shares my, shall we say, profoundly jaded view of the suit side of the record business is probably, like me, thinking it would be a natural.

Not, mind you, with the audiophile audience, of course, but the masses -- who keep getting dragged into this conversation even though I strongly suspect they're more shadow of ghost in this economic machine, as currently configured.
Old 11th April 2014
  #2842
Lives for gear
 
bogosort's Avatar
Quote:
Originally Posted by chrisj View Post
ABX proves to be indeed interesting, but not in the way you're using it, as a weapon and means of belittlement. ABX means any result over 50% indicates a likelihood of SOMETIMES hearing whatever distinction's being tested for, to exact statistical confidence. If you did infinite trials and got 55% confidence you've effectively proved beyond all doubt that a difference was perceptible only some of the time (because at infinite trials, no difference will ALWAYS become 50% confidence)

Or are we going to continue to unscientifically misuse statistics to try to prove a negative?
Ugh, no, that's not how statistics works. Scoring over 50% in an ABX is 100% guaranteed, even if you're randomly guessing -- all you have to do is stop when you've reached your target score. To be meaningful, one must predetermine the number of trials and not throw any results away.

In any case, your final score only tells you the statistical significance of the results -- it represents the probability that you weren't guessing. If you decide to run 16 trials and you get 9 right, the statistical probability that you weren't guessing is 60%. This does not mean that you heard a difference 9 out of 16 times (56% correct): even if you were randomly guessing, you would expect to get 9 out of 16 right 40% of the time.

In other words: You can be guessing 100% of the time and still get a statistical result that suggests you're not. This is why we set significance levels, usually 5% -- if the probability that you were guessing is less than 5%, it is generally considered a significant result. (It should go without saying that if you run 100 such trials, where everyone is randomly guessing, chances are good that at least one will get the same result.)

ABX tests do not prove anything. But if nearly everyone being tested is scoring 95% (not 60%) or better, then we can confidently say that there is an audible difference between the two files. That's it; we cannot extrapolate anything else.
Old 11th April 2014
  #2843
j_j
Lives for gear
Quote:
Originally Posted by chrisj View Post
Anytime you can statistically show through ABX testing above 50% that issues are being heard part of the time, that is worth attending to. 100% confidence is unachievable short of infinite trials, and near-100% confidence is nothing more than a stunt. Give me a break, you guys can't even accept it when someone does get nearly 100% confidence on an ABX!
Not quite. It is possible to get any result in an ABX test randomly, however, the probability falls off rapidly with lots of trials.

For instance, if you run 10 trials, but the answers are random, the chance you get 'n' right is exactly this.

for 0 right, the probability is 0.000977
for 1 right, the probability is 0.009766
for 2 right, the probability is 0.043945
for 3 right, the probability is 0.117188
for 4 right, the probability is 0.205078
for 5 right, the probability is 0.246094
for 6 right, the probability is 0.205078
for 7 right, the probability is 0.117188
for 8 right, the probability is 0.043945
for 9 right, the probability is 0.009766
for 10 right, the probability is 0.000977

All you can ever do is assign a probability that you did (or did not) hear something. In this case, 8, 9 or 10 right would seem to be safe, in that 1 of 20 tests, give or take, would have a false positive. This is called "type 1 error".

It is also possible to have a false negative result. Let's say, for instance, that you are .9 likely to get the right answer. In that case, this is the chances of each of the possible outcomes. Note that the smaller numbers are not really zero, but they are zero to 6 digits, hence the printout. This is also known as "type 2" error. N.B. Some authors conveniently reverse Type 1 and Type 2, but they are both required, mathematically unavoidable potential sources of error.

for 0 right, the probability is 0.000000
for 1 right, the probability is 0.000000
for 2 right, the probability is 0.000000
for 3 right, the probability is 0.000009
for 4 right, the probability is 0.000138
for 5 right, the probability is 0.001488
for 6 right, the probability is 0.011160
for 7 right, the probability is 0.057396
for 8 right, the probability is 0.193710
for 9 right, the probability is 0.387420
for 10 right, the probability is 0.348678

Note here, that for .9 chance of proper ID, the 5% error is around 7. So if you have '7' you would have an indeterminate test in this case, with the hypothesis that you could detect the right answer .9 of the time.

There is never, EVER, an absolute answer out of any proper subjective test, only confidence bounds.

Only probabilities.
Old 11th April 2014
  #2844
Airwindows
 
chrisj's Avatar
Quote:
Originally Posted by bogosort View Post
ABX tests do not prove anything. But if nearly everyone being tested is scoring 95% (not 60%) or better, then we can confidently say that there is an audible difference between the two files. That's it; we cannot extrapolate anything else.
There's no such thing as a binary 'heard/not heard' threshold point. This is your mistake.

If there was such a point, what you say would hold. Since the threshold point is a big gray area, by definition what we get out of ABX testing is a rough metric of how NEAR the threshold point you really are. Assume a very high number of trials, such as (if the threshold was binary) the result would inevitably converge toward either 100% or 50%.

95% and up: very much observing all possible states of the observation.

80%: given enough trials, this says that sometimes the thing is observed and sometimes it is not. When it is observed, it's correctly registered. When it's not, you get guessing.

70%: more guessing, but given enough trials the number of perceptions is still too high to be chance. If it was chance, it would be heading toward 50%. If it persists in being 70% you've got a gray area situation, neither the one or the other.

60%: much more guessing, but hinting at some perceptions coming out. If it was an infinite series this would not occur from pure guessing, only from mostly guessing with the occasional real data skewing things.

Your problem is that you're assuming a binary heard/not heard with no concessions made to the ambiguous nature of perception, where the very existence of blind testing is to illuminate the limits of the ambiguous nature of perception. Very large datasets will not converge only on 50% or 100%. They will seek a level relative to the ambiguity of the perception.

That's not even extrapolation. That comes out of the statistics, and binary heard/not heard is the unwarranted extrapolation. Nothing in the data suggests it.
Old 11th April 2014
  #2845
j_j
Lives for gear
Quote:
Originally Posted by chrisj View Post
There's no such thing as a binary 'heard/not heard' threshold point. This is your mistake.
Actually, you're both wrong here, and making simple mistakes. Please see the post above this and the argument should cease, I think.
Old 12th April 2014
  #2846
Gear Guru
 
UnderTow's Avatar
Quote:
Originally Posted by chrisj View Post
Hey, I never created the files at all.
I now see that you took bandpass' files. What I also see is that the files you posted differ from the files that bandpass provided...

Quote:
The ABX-blind-testing-guys created the files
Ah you are creating camps. Fine. How shall I address your camp? The guys that can't wrap their heads around how (digital) audio and statistics work? Or maybe we can cut the division of people into camps and just address the actual subject at hand.

Quote:
and insisted they were perfect until someone (me, at least) heard differences by cranking my Lavry up (no other changes). The only thing I did was add an extra pair of test files, 5, by running SlewOnly on both 4 files identically. SlewOnly's a freebie audio unit and I linked to it and explained how it works (keeps only the difference between the last sample and the current one: another poster followed up explaining what that does in terms of EQ slope. It wipes out everything but super-highs)
You do realise that your processing of the files completely invalidates any tests? We are discussing delivery formats. That means no ridiculous processing that allows you to play back the files at otherwise insane levels.

Also note that I personally find the test should be done with SoX set to high-quality SRC so as not to effect the audible band. I suspect that the original files were done SoX but without the high-quality setting. I'm waiting for confirmation on this. It might be interesting to find out if people can tell at the lower setting but not at the higher setting but for me that is a different (albeit interesting[1]) discussion. As far as I am concerned, we should only use high-quality settings for these tests. We want to test the audibility of SRC with the best possible settings. We do not want to test whether there are sub-optimal ways to convert files that give audible differences. (That is already a given. There are plenty of examples of bad SRC to give audibly different results).

[1] This is actually an important real world application: If someone tells us they can hear a difference after converting files, it would be useful to be able to point people to possible reasons for this. For instance a different setting in their SRC. Luckily the SRC discussion was already started many many years ago and one of the results of that discussion is the excellent Infinitewave SRC page by Dave Horrocks: http://src.infinitewave.ca/ That is a real useful source of information that can give real results to people by pointing them to better tools without costing anyone any money or limiting their options during production unlike blindly telling people to increase their sample rates with all the extra costs that brings with it and no guaranteed results. I believe that the type of discussion that happened around SRC and the resulting Infinitewave SRC page put real pressure on developers to improve their products. Real science, real measurements, real pressure on companies to improve their products. That is what we should be striving for: Real results based on real measurements and tests to improve the tools we use. This fantasy based quest to buy the most expensive snake oil is not helping anyone but oily salesmen with no integrity (or no clue).

Quote:
We have a more fundamental issue because even if we went to 48K (which I think is well beyond what I need in terms of distinguishing continuous super-highs all the time), that is still entirely in the context of double blind listening. And I've said and continue to say: that which is ABXable is not enough. It is nothing more than the proven minimum you can possibly have. We can't experimentally determine the maximum you need, only come to consensus based on general agreement in all conditions. That is NOT the same as determining the minimum under all conditions…
The blind ABX test is all you reasonably need to know. If people can not detect a difference, there is no audible difference. That's it. (See below for comments on the statistical results).

Quote:
I can establish that there is a threshold past which the files given (the 4 pair and my 5) can always be distinguished, though the only difference is they were filtered and returned to the higher 'carrier' sample rate and bit depth.
I don't agree that this has been determined but more importantly, I would want the test performed with SoX in very high quality mode and with no non-sense processing of the files. Just the original 96 Khz files and a copy that is down-sampled to 48 (or 44.1) Khz and back up to 96 Khz. Can you tell the difference in a blind ABX? If not, there is no significant difference to you. That's all you need to know.

Quote:
Bottom line is it was ABX and you guys are unscientific to invalidate that rather than take it as a data point.
There is no data point from you. Just a big fuzzy unclear experiment that shouldn't be trusted. Just take the original files, convert them with a high quality SRC (SoX in high quality mode is my recommendation as I keep repeating ) down to 44.1 and back up to 96 Khz. Can you tell the difference in a blind ABX test yes or no?

Quote:
So what if I cranked the volume? People do that every day.
You didn't juts crank up the volume. You removed most of the signal making it possible to crank up the volume way way way way way beyond what anyone sane would do with full program material. You created a completely artificial and unrealistic situation to prove your beliefs. You even admitted that you couldn't tell the difference before doing these tricks. Chris, what you are doing is religion, not science.

Quote:
The really big issue here is that everybody in the ABX camp seem to be taking a completely unwarranted illogical jump and trying to use this jump to legislate what people are allowed to have for digital delivery formats, and that's a big problem from where I'm standing, and the reason this discussion is worth having.
You are, again, the one making the unwarranted illogical jumps and, as often before, making stuff up out of thin air. No one is suggesting any legislation.

The ABX test is really simple: It tells you if for delivery files (which means no subsequent processing or other utterly unrealistic mangling of the signals), can anyone tell the difference between between 96 and 96->44.1-> 96 with a high quality SRC. No one has demonstrated that they can. That was the state of affairs before this thread was started and remains the state of affairs until someone demonstrates the contrary.

If anyone "passes" such a test, we can start moving to the next step: Is there a better SRC that does give transparent results? Note that that is the next step that should be performed because the SRC could still be the weak link. We can not directly jump to all sorts of other conclusions before the next step is performed. (Yeah science can be dreary. Unfortunately the results of science are often much more sexy than the grunt work of doing the actual science).

Quote:
The logical mistake is this. You guys are getting actually worked up about any attempt to expand digital audio beyond the minimum proven to be identifiable as a 'fault'.
It is indeed a fault to use twice as much storage and twice as much processing etc to end up with results that are no better. That is just bad engineering not to mention the cases where it is done intentionally to mislead customers...

Quote:
Changes in the super-highs, in the noise floor, ANY change from what you have at 24/96 or 24/192 can and must be considered a fault.
And as soon as anyone can demonstrate that they can perceive a difference when it comes to delivery files, we can start discussing how to address that but we are not there yet. We are exactly where we were before this thread (and many others) started: No one can demonstrate they can hear a difference when using a high quality SRC with the right settings.

Quote:
If it was not different, you wouldn't be able to tell in ABX no matter what you did.
And so far no one has demonstrated that they can.

Quote:
Some of the examples, I couldn't ABX no matter what I did, because to ABX is to conclusively prove beyond ambiguity, 20 out of 20 (or to get a statistical confidence. 20 out of 20 is not 100% confidence. It's just very high)

Nothing about this proves that observations below this 'confidence bar' no longer exist.
Nor does it prove that the opposite even exists. You are, as you have been doing throughout this whole discussion, desperately grasping at straws. There is no evidence that anyone can tell the difference but a vast amount of research to the contrary. The only thing we have is a lot of beliefs that a difference might be audible.

Now here comes the important bit: I'm sure in some cases those beliefs are warranted. There are plenty of bad converters or bad SRC algorithms out there just like there are plenty of bad plugins and consumer players etc. The important question to me is what is the best way forward to achieve good quality audio for everyone? For some people and in some circumstances (let's say someone with bad converters) the quick and dirty solution might just be to record or listen to music at a higher sample rate. Done.

On a superficial level the problem seems fixed for them but is it really? For one they need double the bandwidth, storage and processing (if any processing is done) to achieve the same results as someone that is using quality tools to start with. Secondly, there is reason to believe that if the converters sounds audibly worse at the base rates, there are other issues with the tool (like clocking or analogue stages) that isn't being fixed by simply doubling the sample rate. Thridly, people are attributing the bad results at the base rate to the sample rates instead of rightly pointing at the deficient product and advising everyone to stay away from it.

I would much prefer if the whole industry was pushing to deliver better converters, better plugins, and better consumer DACs that work well at the base rates of 44.1 or 48 Khz. This means a few things: Bad or inexperienced developers/manufacturers/coders can't hide their lack of skills (or too much corner cutting) behind a fake sample rate excuse. Hopefully people will stop using those products in favour of products designed by people with more knowledge and/or integrity. That benefits everyone! Much more than chasing some silly SR fantasy. When audio engineers do that it is like poisoning their own well. It really isn't a good idea IMO.

Also see below about the rare exception of people hearing above 22Khz that might exist.

Quote:
Let's consider a totally failed ABX distinguishing, say, pair 2. You guys are behaving like below a certain point, absolutely no perceptions are accurate and relevant:
Because that is how science deals with these things. And for good reason: You don't know if those few right answers are legitimate correct choices or simply the result of random chance. That is why we need a statistically significant difference or we have to accept that we do not have a conclusive result. We have to reject anything below a certain percentage if we want to trust our test results! That's it. No vague phantasmagorical straw grasping interpretations will do. (Unless they are themselves again appropriately tested. Until then, they are just wishful thinking).

Quote:
Nothing about this PROVES a listener can't legitimately hear a detail once (correctly) in 100 wrong guesses.
Nor does it tell us the opposite, that the listener ever can hear any such difference. That is the problem with the results below a certain percentage; They don't tell us anything as we can not distinguish between pure chance and a few correct choices amongst many incorrect choices. Hence those results are literally meaningless. That is what the stats tell us: How sure are we of the results. A low percentage tells us the results are not trustworthy so we discard them. A high percentage (95% or above) tells us we can be reasonably confident in the results.

That said, when it comes to practical things, any low confidence percentage result does tell us that there are probably a million other things that can be improved and are undeniably audible and will have a direct impact on the listener. Improving these million other things is a much better use of everyone's resources and will give actual audible results to the audience.

This whole "Hi-Rez" dragon chasing is distracting developers and audio engineers etc from real stuff that will, without any doubt, benefit everyone. So yes, this avenue really is bad if you actually care about sound and music!

Quote:
The nature of the lowering confidence more or less establishes that sometimes you legitimately hear a thing and sometimes you don't, and it's a big gray area between the two, not the flipping of a switch. Confidence is statistical.
The results are statistical and include chance guesses so no, you can not at all conclude that anyone is sometimes hearing some difference. You can't actually tell the difference between a few right choices and a few right guesses. That is the reason why something like a 55% result has to be entirely and totally discarded.

There is a whole field on statistics that deals with these things. We can hardly get a simple listening test going on this audio engineering forum so let's just accept the norms for these kind of tests and stop wasting time with you or anyone else re-inventing how scientific research and statistics are done.

Quote:
If the purpose of 96K (never mind 192K ) is to ensure that there is never a dull moment in your music listening then it's gotta be there however your hearing fluctuates. If your attention grabs for a telling detail at 44.1K and it's not there, that's a dull moment. It might have passed by unnoticed, probably would, but at 96K the content would be different and the moment would be different and it would be there to be heard whatever your listening condition or the state of your attention.
All wishful thinking. We have to start at the beginning: Can anyone actually tell the difference to a statistically significant level in a proper ABX test? If we do not have an absolute yes answer to this question, then there really is no point in coming up with fanciful theories like the above.

Quote:
Filtering over 22K? Controversy. Yet it's the same thing.
It is only a controversy because it ties in to people's beliefs that they can hear a difference with higher sample rates. It is not a controversy because anyone is presenting any valid evidence that they can actually hear anything above 22 Khz. So there is no actual controversy. Just people dreaming.

And frankly, even if there are a few kids out there that can hear above 22Khz, I find it extremely unlikely that they care about audio being filtered at 22Khz. I mean, really? That isn't music and in full program material is highly unlikely to be noticeable (even just as an occasional dulling). Either way there are much bigger issues than this but don't get me wrong, if there really is a significant part of the population that can hear above 22 Khz (let's say 1% or more) then I am all for covering their hearing range too by dropping 44.1 Khz and only keeping 48 khz. (Which in the way modern converters and their filters tend to be made, covers up to 23 or 23.5 Khz transparently).

Quote:
Anytime you notice a moment where there should be audio content and it's not there, that is a data point.
No it isn't. Not least because you don't even know if it really happened. See above.

Quote:
ABXing is finding the statistical point where you can conclusively say, ALL these moments are dull, every last one.
No, that isn't what the statistics tell us. They do not relate to how often someone might hear something . The ABX test results should only be read as a positive or a negative. In case of a positive, we know with sufficient confidence that people can tell two files apart. Nothing else.

Quote:
I think many of the people in this thread, if they try, could beat 60% confidence over 20 trials, for the >22K filtering.
I sincerely doubt it but that is the difference between you and me: I am basing my views on all the research that has already been done over the years. Simple Occam's razor really. You are basing your views on wishful thinking (and at least one flawed test). Come up with some real evidence and we can discuss that.

The most striking thing to me is the weird thought processes involved here. Even IF someone can sometimes tell the difference between a 96 K file and one which has everything removed above 22Khz with a high quality filter, how does anyone even think it has any real relation to actual music enjoyment? Higher than 22 Khz content? Are you serious? Or have you simply completely lost the forest for the trees? There are SO many things that actually do affect music and people's experience when listening to music (for instance the performance of our tools at 44.1/48 Khz) that we should be addressing. This "hi-rez" dragon chase really takes away from endeavours that will improve our tools and playback devices. The higher sample rate charade also puts money in the hands of snake oil salesmen which means less money going to developers of good tools and products. It is really a bad thing for music as far as I am concerned.

Alistair
Old 12th April 2014
  #2847
Gear Guru
 
UnderTow's Avatar
Quote:
Originally Posted by bogosort View Post
Ugh, no, that's not how statistics works. Scoring over 50% in an ABX is 100% guaranteed, even if you're randomly guessing -- all you have to do is stop when you've reached your target score. To be meaningful, one must predetermine the number of trials and not throw any results away.
And we shouldn't be looking at the results until the end of the test as that can bias us. (Many tools show the Guessing probability after each response).

Alistair
Old 12th April 2014
  #2848
Gear Guru
 
UnderTow's Avatar
Quote:
Originally Posted by theblue1 View Post
It may present an interesting sidelight on the issue of why some plugins are reputed to perform better at double or quad rates... the speculation in such cases is often that the coder didn't check for processes that had created out-of-band content. Sometimes people wonder how a coder could make such a mistake...
Heh. The other issue is that they did think of out of band content but didn't implement the filters well. Sometimes they are too low and thus audible or they are not steep enough so they cause aliasing oir even worse, just sound wrong. Either way, I think it would be best for the whole audio engineering community if the default response to someone complaining about a tool not sounding as good at the base rates would be people telling others not to use that tool because it is well known not to have good filters and/or not perform well at the base rate. Or even better, let the users petition the developers to invest time into researching and implementing proper filtering.

Improving the tools should be the goal. Not just putting band-aids (with all the resulting extra resource costs), on sub-optimal products.

Alistair
Old 12th April 2014
  #2849
j_j
Lives for gear
Apparently since the various squabbling over simple statistics is from people who won't even read the simple answer above, probably because they block all contrary input, I give up.
Old 12th April 2014
  #2850
Gear Guru
 
UnderTow's Avatar
Quote:
Originally Posted by j_j View Post
Apparently since the various squabbling over simple statistics is from people who won't even read the simple answer above, probably because they block all contrary input, I give up.
Well thanks for participating for as long as you did. It is always interesting to read your posts!

Alistair
Topic:
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump