[Dobrescu George]

   
Look forward to your reporting of results, could be very interesting. Of course, your simply asking people if they heard a difference between 2 tracks after listening to them is probably a simplification of your methodology, hopefully? Regarding 3, have you take any frequency response measurements on the 2? Could be a problem with one, I had a sample X5, and it sounded pretty darn near identical to the Clip+, but there could be a problem with one of yours (or one of mine) that led to differences in results between your testing and mine. And maybe you've got better ears.

Methodology and other details cannot be made public until I finish the project, which will last at least 2 more years until it comes to it's results.
 
 
About 3, it is interesting that some people had not noticed the differences. 
 
For me it was audible, like a difference in signature overall, not just minor differences. BTW, I am using X5II not first gen x5 at the moment, the sound changed in second version, for the good. It responds better to treble, better transients, and better dynamics. Maybe the very efficient amp stage made the difference. Then again, all my music sources have their sound, if I were to say, X5II is the best, while others are all off. Not by much, but it is audible to the average person, without using expensive headphones or anything. Every single device had a signature of it's own. Or maybe I am too picky, and use music that just over-accentuates these differences. 

 

[sonitus mirus]

  Methodology and other details cannot be made public until I finish the project, which will last at least 2 more years until it comes to it's results.
 
 
About 3, it is interesting that some people had not noticed the differences. 
 
For me it was audible, like a difference in signature overall, not just minor differences. BTW, I am using X5II not first gen x5 at the moment, the sound changed in second version, for the good. It responds better to treble, better transients, and better dynamics. Maybe the very efficient amp stage made the difference. Then again, all my music sources have their sound, if I were to say, X5II is the best, while others are all off. Not by much, but it is audible to the average person, without using expensive headphones or anything. Every single device had a signature of it's own. Or maybe I am too picky, and use music that just over-accentuates these differences. 

 
I certainly hope that you provide all of the details about how the test files and equipment were volume matched, how the switching was handled, and that reasonable assurances can be shown that bias was removed.
 
Is it a trade secret to tell us a little bit about how some of this is being accomplished?  
 
I would be disappointed to learn that the results will just be another one of those that we simply have to listen for ourselves to hear these obvious differences.  At this point, I feel obligated to tell you that something seems to be amiss with your testing setup.  I mean, gravity is not being reversed or anything like that, but your results are not in line with the few published test results I have been able to read, and they seem to be leaning toward the realm of the marketing and business-oriented findings on the matter. 
 
The baseline should be that nobody has provided any verifiable proof that SACD is any better than Red Book.  There was the Meridian-backed AES convention paper that was largely discredited and was never eligible to be peer reviewed.  That really doesn't hold much water at this point.
 
https://secure.aes.org/forum/pubs/conventions/?ID=416
 
And we have Amir, the champion of all things high definition audio, that seems to be on a crusade to prove that a difference can be heard.  His claims seem valid, but it is difficult to trust his results based on his open agenda, and there really isn't any evidence that his results can be duplicated by others.
 
http://www.whatsbestforum.com/showthread.php?15255-Conclusive-quot-Proof-quot-that-higher-resolution-audio-sounds-different
 
In other words, you should be very concerned when you can easily hear differences when a properly controlled ABX test is performed.  

 

[Brooko]

Oh come on George - that's being evasive.
 
Simple methodology without giving away too much detail.  You can't simply say "Multiple subject blind testing"
 
So:
 

• Was testing done with music from the same master file?  What bit-rates did you use for comparison?
• When you tested - did you use a high-res master, then create a copy - resample and dither to selected lossy bit-rate - then upsample again for the actual tests? 
• Did you volume match?  How?
• Who controlled the actual blind test?  What equipment and software was used?  Was it a true ABX?
• How many iterations?  What do you consider statistically relevant result?
• IMPORTANT - was each test on the same equipment, same person, same track, same source, same volume etc - with only change being the container / bit-rate?

 
And just for interests sake - how about uploading two files you've used so we can have a look at them?  Then we can also add to your data 

 

[rsaavedra]

Originally Posted by Dobrescu George /img/forum/go_quote.gif
This is false, mathematically, I am studying in university and it is an advanced research going on about this, and the result is that high bitrate clearly is better than CD quality. How audible it is, is another thing, but in theory the differences are there, and audible. My PhD is based upon this, and it is supported with mathematical and other types of evidence.

Well well, you here say that "the result is that..." bla bla bla, so your "it is an advanced research" has already some results, but you later say can't show any evidence until two years? Sorry man, I actually don't believe you are doing any PhD. Quite a few things in your post smell a bit fishy to me. Where are you pursuing this degree by the way? And a PhD in what exactly?
 
For one thing, you could easily post a link to your PhD student research page, most if not all PhD students do have such, and is not an external blog, it's usually a page within the university webpages, in fact within the academic department webpages, sections "PhD students and research" or the likes. As PhD candidates, they explain there clearly who the are, what they've done, what their interests are, and what research they do, links to related research and work, papers they've already published, or research in which they've participated, who their advisor is, and so fort. I'd appreciate if you shared any of this. Could you?
 
One other rather fishy thing at least for me from your post, you say "please do not disninform people." Well, claiming that you have "results" and the truth in your hands, and then claiming that you can't show any evidence ***IS*** misinforming, and shamefully so, especially for any PhD candidate. It would be just bluff. In fact, even if you already published any results from a successful PhD thesis, you still couldn't claim that's the ultimate truth at all or that's the way things are. Other researchers would need to corroborate and replicate, after all any research can have flaws, a PhD candidate ought to know that. Plus, just as for non-researchers or PhD candidates, there is always the not-insignificant number of fraudulent researchers out there, don't you know?
 
In any case, the actual work and publications from a researcher are what speak on behalf of their research, not posts on forums saying they are Phd candidates (or even PhDs already) and have worked on so and so but can't show any actual results. That is as lame as can be for a real researcher, absolute bull.

So I would ask you to clean your act a little. Show at least valid evidence that you are indeed pursuing a PhD, if it's true, but most importantly, you please do stop misinforming people.

 

[reginalb]

  Methodology and other details cannot be made public until I finish the project, which will last at least 2 more years until it comes to it's results.
 
 
About 3, it is interesting that some people had not noticed the differences. 
 
For me it was audible, like a difference in signature overall, not just minor differences. BTW, I am using X5II not first gen x5 at the moment, the sound changed in second version, for the good. It responds better to treble, better transients, and better dynamics. Maybe the very efficient amp stage made the difference. Then again, all my music sources have their sound, if I were to say, X5II is the best, while others are all off. Not by much, but it is audible to the average person, without using expensive headphones or anything. Every single device had a signature of it's own. Or maybe I am too picky, and use music that just over-accentuates these differences. 

 
My own testing was with the first X5, haven't used the second. 
 

  Well well, you here say that "the result is that..." bla bla bla, so your "research" has already some results, but you later say can't show any evidence until two years? Sorry man, I actually don't believe you are doing any PhD. Quite a few things in your post smell a bit fishy to me. Where are you pursuing this degree by the way?
 
For one thing, you could easily post a link to your PhD student research page, most if not all PhD students do have such, and is not an external blog, it's usually a page within the university webpages, were they, as PhD candidates, explain clearly what their interests are and what research they do, links to related research and work, papers they've already published, or research in which they've participated, who their advisor is and so fort. I'd appreciate if you shared any of this. Could you?
 
One other rather fishy thing at least for me from your post, you say "please do not disninform people." Well, claiming that you have "results" and the truth in your hands, and then claiming that you can't show any evidence ***IS*** misinforming, and shamefully so, especially for any PhD candidate. In fact, even if you already published any results from a successful PhD thesis, you still couldn't claim that's the ultimate truth at all. Other researchers would need to corroborate and replicate, after all any research can have flaws, a PhD candidate ought to know that. Plus, just as for non-researchers or PhD candidates, there is always the not-insignificant number of fraudulent researchers out there. In any case, the work and publications from a researcher are what speak on behalf of their research, not posts on forums saying they are Phd candidates and have worked on so and so but can't show any actual results.

So I would ask you to clear your act a little. Show at least valid evidence that you are indeed pursuing a PhD, if it's true, but most importantly, you please do stop misinforming people.

 
I mean, I understand where he's coming from. I did some research in grad school, and had some situations where all I could give was some general information. Now, these weren't experiments, they were econometrics, and I might not have done a full regression or whatever. So I might have an idea what the data set was going to tell me, but couldn't really give detail, because I didn't have it yet. Also, there is competition in academia, for better or worse, no reason to be overly hostile. I will remain skeptical until research with valid methodology is published to contradict my skepticism, of course, but no reason to be hostile.

 

[rsaavedra]

  I mean, I understand where he's coming from. I did some research in grad school, and had some situations where all I could give was some general information. Now, these weren't experiments, they were econometrics, and I might not have done a full regression or whatever. So I might have an idea what the data set was going to tell me, but couldn't really give detail, because I didn't have it yet. Also, there is competition in academia, for better or worse, no reason to be overly hostile. I will remain skeptical until research with valid methodology is published to contradict my skepticism, of course, but no reason to be hostile.

I did research in grad school as well and at work. And I *do not* understand where he's coming from. And this is the section about "Sound Science" after all, and the title of this thread can clearly suggest why I might be all out hostile against what seems to me a lot of plain bull.

 

[reginalb]

  I did research in grad school as well and at work. And I *do not* understand where he's coming from. And this is the section about "Sound Science" after all, and the title of this thread can clearly suggest why I might be all out hostile against what seems to me a lot of plain bull.

 
Perhaps English isn't his first language, and he's unable to really articulate the what and why he can't talk about it more. Sure, it might be bull. And I'll await publication before I allow it to sway any of my own opinions. But it is sound "science" not sound "every question has been answered definitively," so I will remain skeptical, sure, but no reason to get hostile.  

 

[money4me247]

  I had not read the entire thread, but I had read that page. I agree with most statements, but now with:
 
7. Audiophile higher bit audio formats are not likely to offer audible improvements over compact disc 16 bit/44.1 kHz quality.
 
This is false, mathematically, I am studying in university and it is an advanced research going on about this, and the result is that high bitrate clearly is better than CD quality. How audible it is, is another thing, but in theory the differences are there, and audible. My PhD is based upon this, and it is supported with mathematical and other types of evidence.

Can't be false mathematically since Nyquist's sampling theorem already proves mathematically that waveforms can be recreated perfectly using a sampling rate twice that of the frequency response range. No information will be lost if the signal is sampled at the Nyquist frequency and no information will be gained using faster sampling rates.
 
Mathematics below:
 
Let 

 be the spectrum of 

  Then

since 

 is assumed to be zero outside the band 

. If we let

where n is any positive or negative integer, we obtain

 

 

[mlxx]

  Can't be false mathematically since Nyquist's sampling theorem already proves mathematically that waveforms can be recreated perfectly using a sampling rate twice that of the frequency response range. No information will be lost if the signal is sampled at the Nyquist frequency and no information will be gained using faster sampling rates.

 
Yes the maths aren't wrong but this can never work in practice because you can never take "exact" samples. By this I mean that you are sampling at 16bits (or 24bits or whatever) which is never exact, so it will never be possible to recreate the waveform "perfectly" in practice. In which case, increasing the bit depth or the sampling rate will make a difference in regards to producing a more accurate waveform. So in practice, 24bits will be more accurate than 16bits, 16/88 will be more accurate than 16/44 etc. If this is audible or not is a another matter though.

 

[RRod]

   If this is audible or not is a another matter though.

 
Isn't that the whole point, though?

 

[mlxx]

   
Isn't that the whole point, though?

 
My point was that you can never reconstruct the waveform perfectly in the real world because you can never sample with infinite precision. So there will always be some better way of making it more accurate because it will never be perfect. I haven't bothered to test the audible differences myself, if someone can tell the difference or not... I don't know and I don't care... quite happy with the CD stuff myself.

 

[castleofargh]

-step one find kids with hearing up to 25khz
-step 2 make them blind test 16/44 vs 24/96 files that have fullscale 24khz tones.
 
highres sounds audibly different= proved!!!!!
 
 
alternative, do the same stupid stuff with bit depth and -120db signal pushed real loud. night and day difference!!!! boom head shot!
 
 
I've proved that hires is better twice already, wake up sound science, highres is so much better for the guy recording dolphins from too far away!!!! so obviously we all need to have it in our homes.
or are you saying that you hate dolphins and want to hurt them? 

 

[Dobrescu George]

  Oh come on George - that's being evasive.
 
Simple methodology without giving away too much detail.  You can't simply say "Multiple subject blind testing"
 
So:
 

• Was testing done with music from the same master file?  What bit-rates did you use for comparison?
• When you tested - did you use a high-res master, then create a copy - resample and dither to selected lossy bit-rate - then upsample again for the actual tests? 
• Did you volume match?  How?
• Who controlled the actual blind test?  What equipment and software was used?  Was it a true ABX?
• How many iterations?  What do you consider statistically relevant result?
• IMPORTANT - was each test on the same equipment, same person, same track, same source, same volume etc - with only change being the container / bit-rate?

 
And just for interests sake - how about uploading two files you've used so we can have a look at them?  Then we can also add to your data 

 
1. Testing done from the same master file, Redbook CD file(s), using the original file and the resampled file.
2. Simply the original and my version of it. 
3. Volume matched.
6. Yes, everything remained the same, sometimes even trying to figure out if the position of body or mental state of subjects affects the results. There will be testing until all data that could had been subjective thinking of the subjects is eliminated. And then will be more testing, just because the subjects must vary in every possible way.  
 
Staff where you present your PhDs are much worse than soundscience when it comes to being skeptic. Until your data cannot be countered, you are sent back to testing. 
 
For this project, it is very early, I just wanted to point out that bit rate is not all the same. We are trying to interpolate better using more samples, not obtain ultrasonic frequencies, the main problem is to also make it audibly better. Mathematical differences are easy to obtain, but making it sound better for human hearing is another thing.
 
Files will be put on the project's blog when I have finished completing the copyrights, but it will take a while. The project will also be made public when in late stages. 

 

[RRod]

   
My point was that you can never reconstruct the waveform perfectly in the real world because you can never sample with infinite precision. So there will always be some better way of making it more accurate because it will never be perfect. I haven't bothered to test the audible differences myself, if someone can tell the difference or not... I don't know and I don't care... quite happy with the CD stuff myself.

 
I think we all agree with "A>B therefore A>B", the real question is defining ">". If I want to play back a volcano eruption, then our concept of what is "better" is much different than for human hearing. But I recall a line from a DSP book that went something like "DSP should be tailored to the application." Simply increasing bits and samples because "more is better" is the exact opposite of this mantra.

 

[RRod]

Originally Posted by Dobrescu George /img/forum/go_quote.gif
 
Staff where you present your PhDs are much worse than soundscience when it comes to being skeptic. Until your data cannot be countered, you are sent back to testing. 
 
For this project, it is very early, I just wanted to point out that bit rate is not all the same. We are trying to interpolate better using more samples, not obtain ultrasonic frequencies, the main problem is to also make it audibly better. Mathematical differences are easy to obtain, but making it sound better for human hearing is another thing.

 
It will be interesting to see the results. Hopefully you can understand some of the skepticism on here: people haven't been able to reliably show the ability to discern improvements from hi-res, and that's even with hi-res content actually in the track. It's thus hard to accept that people would be able to discern improvement from upsampled Redbook, unless something like an audible filter, extra distortion, boutique dither, or switching issue is the cause.