LHC Posted July 5, 2016 Posted July 5, 2016 (edited) http://www.aes.org/e-lib/browse.cfm?elib=18296 free download Peer reviewed paper. Not workshop. Meta analysis indicating the difference between Redbook and hi-res is audible. and: http://www.prosoundweb.com/article/p...solution_audio “Our study finds high-resolution audio has a small but important advantage in its quality of reproduction over standard audio content. Trained listeners could distinguish between the two formats around sixty percent of the time.†Dr Mark Waldrep has a balanced write up about the result of this literature survey here http://www.realhd-audio.com/?p=5755 Whilst the finding of the survey is encouraging, further careful work is needed to narrow down what caused the advantage in quality. So Waldrep's issue about provenance is quite important if future tests are to show that ultrahigh frequencies are responsible for these quality improvements. In other words the case about what people can or cannot hear is far from closed, and much more rigorous research is needed. Edited July 5, 2016 by LHC
JSmith Posted July 5, 2016 Posted July 5, 2016 I am not satisfied with digital audio nor the mathematics and theory behind it. Please explain your issues with the mathematics. JSmith 1
Newman Posted July 5, 2016 Posted July 5, 2016 ...@@eltech, You should look into it further. All the explanations you seek are there, and not particularly controversial. Agreed. But despite numerous such recommendations, he refuses, then simply comes out in the next thread and repeats (as facts, not opinions) all the factually incorrect misunderstandings of everything that is simply digital electronics 101. And that's when one starts to question a person's intent.
Newman Posted July 5, 2016 Posted July 5, 2016 ... I perceive analogue to out perform digital based on what I've heard in spite of its numerous technical documented and acknowledged flaws. So I approach the topic based on the criterion of personal perception of sound quality. I have to assume that digital sampling is imperfect, flawed and has room for improvement in spite of its theoretical "perfection". ....and here is the clue to that intent. Post-rationalisation. If only you were paying attention to the discussions of sighted listening and what happens, your above-mentioned basis, approach, and 'have to assume' could all be dropped, because the explanations of your listening experiences are far more mundane: you can't trust your ears when your mind is allowed to override them. Only when we 'let go' of the false reality that sighted listening tells us is true, can we then let go of the edifice of false science and false attributions between cause and effect, that we create to justify that false reality. Some people find this step excruciatingly difficult. I don't blame them.
LHC Posted July 5, 2016 Posted July 5, 2016 understanding how it performs and how it is perceived to perform are separate things. I perceive analogue to out perform digital based on what I've heard in spite of its numerous technical documented and acknowledged flaws. So I approach the topic based on the criterion of personal perception of sound quality. I have to assume that digital sampling is imperfect, flawed and has room for improvement in spite of its theoretical "perfection". Indeed you may be correct about inaudible higher frequencies, or not. But I still search for the reasons for what I hear and perceive. I am not satisfied with digital audio nor the mathematics and theory behind it. I didn't want to, and don't want to distract this thread from its original topic, but the only way for me to further state my position was to say the above. Good on you eltech, I support your stance on 'trusting your ears'. No one should tell you what you like and what you don't like. You are not deluding yourself. Having said that, there are merits in blinding listening tests, so that is something worth trying if you haven't done so already. This is a long thread so it is worth recapping some of what I found earlier. Using higher sampling rate like 192kHz is not about trying to hear any particular sound frequencies above the human threshold of 21kHz. Not directly anyway. Using higher sampling rate like 192kHz is not about trying to find flaws or loopholes in sampling theory. There are certainly issues in its real world implementations, but these have been well studied. So presumably that side is under control. Why we need 192kHz is because human hearing can 'resolve' sound peaks that are very close to each other in time. A distinguished physics professor Kunchur has published test results showing our time coherence threshold can be as low as 6 microsecond. These works have been scientifically scrutinised and published in prestigious peer-reviewed journals. As far as I know no one has published papers that directly refute or discredit Kunchur's findings. So it is accepted scientific knowledge at this stage. Based on Kunchur's findings, engineers at Yamaha argued that higher sampling rates up to 192kHz is needed to reproduce music that have such high resolution in the time domain. Their white paper is here and worth reading (http://www.yamahaproaudio.com/global/en/training_support/selftraining/audio_quality/chapter4/02_audio_universe/) Very smart, knowledgeable and respected engineers like Bob Stuart has also buy into the importance of resolution in time. His interpretation of that is given here http://www.realhd-audio.com/?p=5639 and elsewhere. Finally it is also true that high sampling rate like 192kHz allows engineers to process digital signals better. But the key is what is the end goal? I believe the end goal should include the exploitation of our very low time coherence threshold (as discussed by Kunchur/Yamaha/Stuart) I agree with Dave's comment that doing more reading about these subjects is good. But do maintain an open mind and listen to your ears too.
Happy Posted July 5, 2016 Posted July 5, 2016 i too would advocate trust your ear thing. but trusting your ear is one thing, and not accepting scientific findings is quite another thing. i simply don't see why one has to refuse to accept them.
Newman Posted July 5, 2016 Posted July 5, 2016 (edited) @@LHC, And who, pray tell, has told anyone what they should or shouldn't like? The discussion at the moment is about whether CD level resolution has inherent audible flaws when used for distribution as final product. eltech has been repeatedly posting a stream of statements on that topic that are 100% incorrect, then ignoring corrections, then reposting in every thread where he sees relevance. If anyone is telling others what they should or shouldn't like, it is eltech i.e. we shouldn't like CD quality audio because it can't represent an audible music signal accurately. Not enough sample points. By your silence in response to his claims, I take it that you agree with them? Edited July 5, 2016 by Newman
Nada Posted July 6, 2016 Posted July 6, 2016 ...... This is a long thread so it is worth recapping some of what I found earlier...... Why we need 192kHz is because human hearing can 'resolve' sound peaks that are very close to each other in time. A distinguished physics professor Kunchur has published test results showing our time coherence threshold can be as low as 6 microsecond. These works have been scientifically scrutinised and published in prestigious peer-reviewed journals. As far as I know no one has published papers that directly refute or discredit Kunchur's findings. So it is accepted scientific knowledge at this stage. Based on Kunchur's findings, engineers at Yamaha argued that higher sampling rates up to 192kHz is needed to reproduce music that have such high resolution in the time domain. Their white paper is here and worth reading (http://www.yamahaproaudio.com/global/en/training_support/selftraining/audio_quality/chapter4/02_audio_universe/) Very smart, knowledgeable and respected engineers like Bob Stuart has also buy into the importance of resolution in time. His interpretation of that is given here http://www.realhd-audio.com/?p=5639 and elsewhere. Finally it is also true that high sampling rate like 192kHz allows engineers to process digital signals better. But the key is what is the end goal? I believe the end goal should include the exploitation of our very low time coherence threshold (as discussed by Kunchur/Yamaha/Stuart) Most of the above is simply incorrect. Repeating nonsense doesnt make it any truer.
davewantsmoore Posted July 6, 2016 Posted July 6, 2016 Whilst the finding of the survey is encouraging, further careful work is needed to narrow down what caused the advantage in quality. Indeed .... however the furfie perpetuated widely is that this work has not been ongoing for a long long time (which it has). There is no controversy about AudioA sounding better than AudioB. However there are incredibly large powers that be who are desperate to sell you their back catalogue in a new format. Wouldn't people be upset if we all repurchased a new format, and then discovered that the technique they had used to "improve the quality" could have been delivered by other less costly ways (lock-in, compatibility, consumers $) No one should tell you what you like and what you don't like. Do you think anything even close to that has happened in this thread? (I'd be fairly perplexed if you said yes) This is not about what one likes.... but the explanations given for how something works. It's like if I said I like chocolate ice cream (but I really don't like vanilla) .... If I started explaining to people that the reason vanilla tastes bad is because they put dog droppings in it. Then this would be clearly incorrect (or able to be demonstrated one way or another at least). The fact that a number of people chime in that they do (or don't) also like vanilla icecream has absolutely no bearing on whether it does (or doesn't) contain dog droppings. Why we need 192kHz is because human hearing can 'resolve' sound peaks that are very close to each other in time. A distinguished physics professor Kunchur has published test results showing our time coherence threshold can be as low as 6 microsecond. These works have been scientifically scrutinised and published in prestigious peer-reviewed journals. As far as I know no one has published papers that directly refute or discredit Kunchur's findings. So it is accepted scientific knowledge at this stage. The practical application of this work has been misunderstood by many. It IS correct that these levels of delay can be distinguished, and this was already known before that paper (the paper just did a good job of demonstrating it) .... however the results of the paper do not directly support a need to distribute audio with a high sampling rate. You can arrange the delays being discussed using any sampling rate. The results DO however, support the need to resample audio (up and down) very very carefully, as you can harm the audio (and "blur" those mentioned time features) by doing to.... this perhaps supports the idea that you should probably not resample audio at all (to be safe) .... OR use a very high quality rendering method if you were going to resample it, and not let it be inadvertently resampled by something else (eg. your DAC internal resampling). This is what MQA (and some others for example xxhighend, hqplayer) are doing. Using large amounts of computer power to resample audio in a way which doesn't harm it. It is why DSD formats can sound superior (because it saves the DAC from having to resample it internally) Based on Kunchur's findings, engineers at Yamaha argued that higher sampling rates up to 192kHz is needed to reproduce music that have such high resolution in the time domain. Their white paper is here and worth reading (http://www.yamahaproaudio.com/global/en/training_support/selftraining/audio_quality/chapter4/02_audio_universe/) Can you explain their reasoning? They only seems to discuss how we can hear such short delays (which is not controversial). They don't appear to make a (even a brief) case for digital audio format [which isn't surprising] Finally it is also true that high sampling rate like 192kHz allows engineers to process digital signals better. But the key is what is the end goal? I believe the end goal should include the exploitation of our very low time coherence threshold (as discussed by Kunchur/Yamaha/Stuart) A high rate isn't necessarily required to do that. If you want to resample audio (which almost always happens), you need to be particularly careful otherwise you may harm the audio..... otherwise either leave it in the original rate, or render it very carefully. I agree with Dave's comment that doing more reading about these subjects is good. But do maintain an open mind and listen to your ears too. The thing to be aware of is that just because you hear something, doesn't mean you understand exactly what caused it (digital audio is quite complex, certainly more so than has been unpacked in any of these threads which are normally dealing with simple/fundamental concepts of "how it works" (or doesn't). "confirmation bias" from hearing A sound better than B .... is a very powerful thing. It happens all the time in life. A did sound better than B .... it's just that the belief about WHY is misfounded. Leaning more about digital audio may help to see that. No one should tell you what you like and what you don't like. 1
Nada Posted July 6, 2016 Posted July 6, 2016 (edited) 192kHz sampling has, in my view some clear benefits: 1. If a DAC has poor or no upsampling 192kHz can sound better then 44.1 (but most audiophile DACs arent handicapped in this way) 2. 192kHz is profitable for "HD" providers (but smart audiophiles can just rip CDs and save the money and storage hassles) 3. placebo response allows 192 to sound better, after bigger is better right (unless youve seen through that myth) 4. If downsampling SRC in a studio are poor then 192kHz might sound better (unlikely) 5. it gives some irrational vinyl trolls or retailers ammunition to push their plastic agenda Edited July 6, 2016 by Nada 1
Newman Posted July 6, 2016 Posted July 6, 2016 It's like if I said I like chocolate ice cream (but I really don't like vanilla) .... If I started explaining to people that the reason vanilla tastes bad is because they put dog droppings in it. Then this would be clearly incorrect (or able to be demonstrated one way or another at least). The fact that a number of people chime in that they do (or don't) also like vanilla icecream has absolutely no bearing on whether it does (or doesn't) contain dog droppings. An excellent analogy for the claims that I have been trying to address.
davewantsmoore Posted July 6, 2016 Posted July 6, 2016 192kHz sampling has, in my view some clear benefits: 1. If a DAC has poor or no upsampling 192kHz can sound better then 44.1 (but most audiophile DACs arent handicapped in this way) Indeed.... but this doesn't lead to a requirement to sample and distribute audio at high rates. Players can over sample the audio where the DAC requires it. If the DAC requires it, but doens't implement it, then this really qualifies as poorly designed equipment though. 2. 192kHz is profitable for "HD" providers 3. placebo response allows 192 to sound better Consumers will and should wise up to this. A more reliable way to indicate quality could be useful (although is also fraught with danger) 4. If downsampling SRC in a studio are poor then 192kHz might sound better (unlikely) Perhaps not as unlikely as you think. There are heaps of papers, books and patents, showing how this can be not so good for audio. Much of the MQA technical info is about this.... It is essentially the main reason why they are advocating for high rates when the original exists in these rates (because lower rates are likely "damaged").
eltech Posted July 6, 2016 Posted July 6, 2016 @@davewantsmoore Can you tell me what your position is on this topic of 192khz and MQN?
eltech Posted July 6, 2016 Posted July 6, 2016 Just to share an observation I have made. When I digitally record a vinyl LP the crackles and pops don’t sound right when played back. They sound like the high frequency content of the crackles and pops has a low pass filter on them, they sound dull, and they sound slow - is the best way I can describe it. When playing the vinyl record directly through the amp and speakers the same crackles and pops have a lot of high frequency content that seems to go sky high and they sound fast, and less intrusive. I have found that higher sampling rates do a better job of capturing and playing back the crackles and pops, by that I mean they sound more like listening directly to the record. I have an ESI Juli@ soundcard. I don't know if its AD converter is supposed to be good though it can record 192khz. My personal observation is one of the reasons I think higher sampling rates are advantageous.
davewantsmoore Posted July 6, 2016 Posted July 6, 2016 I'm not sure what is MQN? Also can you ask a much more specific question? ..... as giving short answers to such an open ended question invites assumption or misinterpretation. The topic is infinitely more complex than "Yes or No" to is a high sampling rate worthwhile. If you're looking for a "simple truth" .... the most relevant one I think is that the "sampling rate does NOT allow you to necessarily infer anything about the quality of the audio". High rates potentially do two main things: Let your converter work differently (better?!) to when it was working with lower rate audio ..... over sampling can (and does) potentially achieve the same benefits (the quality of the resampling is important however) Avoid resampling all together - by just staying in the original rate (which might be a high rate - that was used to avoid the first issue)
Guest Eggcup The Daft Posted July 6, 2016 Posted July 6, 2016 (edited) Why we need 192kHz is because human hearing can 'resolve' sound peaks that are very close to each other in time. A distinguished physics professor Kunchur has published test results showing our time coherence threshold can be as low as 6 microsecond. These works have been scientifically scrutinised and published in prestigious peer-reviewed journals. As far as I know no one has published papers that directly refute or discredit Kunchur's findings. So it is accepted scientific knowledge at this stage. Kunchur is distinguished in other areas of physics, but is controversial in the audio field. His experiment was simple - he moved one of a pair of speakers around until a listener at a fixed point heard a change when playing back an impulse. He has had to spend a lot of time defending his result from criticism; it is certainly not accepted science, and attempts to reproduce the result have actually given, among other things,a finer threshold result than his original experiment. As far as I can see, this experiment in the form described only proves that moving a speaker a small distance, in a particular environment, caused a change in sound that could be detected. There could be other reasons why the test subjects detected the move. I'll leave it to the experts to say more than that. But I'd feel more comfortable with the impulse response explanation if the experiment was done using a controlled electronic delay rather than moving the speaker. Edit - "moved around" is an unflattering description: the speaker was moved forwards to decrease the distance between it and the listener and therefore change the time the impulse sound took to reach the ears. Edited July 6, 2016 by Eggcup The Daft
davewantsmoore Posted July 6, 2016 Posted July 6, 2016 My personal observation is one of the reasons I think higher sampling rates are advantageous. I understand. You got a better result with 192khz. You can't say why though. There may be something about how your card, or computer works with the different sampling rates, that is nothing to do with general "flaws" in lower sampling rates. Your AD converter doesn't run at 192khz. Is uses a rate of 24,000khz (24mhz) ...... when it reduces (decimation) the rate back to 48khz, or 192khz, or whatever you're asking for .... then perhaps it does something differently depending on which rate you choose. The decimation may be harming the quality when going to a lower rate ..... but this isn't becuase the "lower rate isn't capable of representing your audio properly" ..... it is because your device is not doing a good job of it. The result may be exactly the same to you (ie. the higher rate works better) ...... but it isn't correct to make sweeping statements about "high rates, yes or no". 2
davewantsmoore Posted July 6, 2016 Posted July 6, 2016 Kunchur is distinguished in other areas of physics, but is controversial in the audio field. His experiment was simple - he moved one of a pair of speakers around until a listener at a fixed point heard a change when playing back an impulse. He has had to spend a lot of time defending his result from criticism I think we need to be clear here. His RESULTS are non-controversial. Even finding a shorter time duration probably would not be particularly controversial. It is the CONCLUSION which he draws about his results (and it's application to digital audio), which is off. Essentially he uses his results to support the need for high rates, and that is not justified. A finer point which gets missed however, is that let's say you have audio which contains these time durations (it did not need to be "192khz" to contain these) .... and you resample it. It is possible (somewhat likely even) that you have distorted the audio when viewed on the time axis (and you no longer have these time features, which he results show are audible) .... high rates aren't a panacea for this. If I record audio in 192khz, and then resample it to DXD (352.8khz) for example. I may also have wrecked it.
Guest Eggcup The Daft Posted July 6, 2016 Posted July 6, 2016 I think we need to be clear here. His RESULTS are non-controversial. Even finding a shorter time duration probably would not be particularly controversial. Can I come back, just to be clear? There is no other possible reason for hearing a difference in sound, for example a phase shift, a mechanical change resulting from the speaker movement, or anything environmental?
eltech Posted July 6, 2016 Posted July 6, 2016 (edited) I'm not sure what is MQN? Sorry, MQN is a playback software, and I made a typo and meant to ask you about MQA. Edited July 6, 2016 by eltech
Nada Posted July 6, 2016 Posted July 6, 2016 Just to share an observation I have made. When I digitally record a vinyl LP the crackles and pops don’t sound right when played back. They sound like the high frequency content of the crackles and pops has a low pass filter on them, they sound dull, and they sound slow - is the best way I can describe it. When playing the vinyl record directly through the amp and speakers the same crackles and pops have a lot of high frequency content that seems to go sky high and they sound fast, and less intrusive. I have found that higher sampling rates do a better job of capturing and playing back the crackles and pops, by that I mean they sound more like listening directly to the record. I have an ESI Juli@ soundcard. I don't know if its AD converter is supposed to be good though it can record 192khz. My personal observation is one of the reasons I think higher sampling rates are advantageous. Do you do stand up comedy too? Your hilarious. Let me know when youve got a gig on please.
davewantsmoore Posted July 6, 2016 Posted July 6, 2016 Can I come back, just to be clear? There is no other possible reason for hearing a difference in sound, for example a phase shift, a mechanical change resulting from the speaker movement, or anything environmental? Of course, any number of things could possibly have an affect .... but by measuring the sound, it is easy enough to determine that nothing except the time delay (ie. the phase) has changed. However, it is EXPECTED that he would find the results he did. There's isn't any big obvious reason to pick apart his testing. Other people can do the experiment themselves, and confirm the results (or not). You could use other methods for inducing the delay (like an electronic delay, for example) ..... however there isn't any reason at all to expect that would change the outcome. All in all, it's a fairly simple and easy to control experiment. He just says (or at least implies) the results MEAN something they don't. Sorry, MQN is a playback software Heh, that's all I could come up with too. ... So what's the question? (I've written a lot in the MQA thread - have you read that?). Do you do stand up comedy too? Your hilarious. Let me know when youve got a gig on please. I'm not quite sure which bit you think is "comedy" .... but just taking his reported result on it's own - what is so funny? His result is the result..... and simply reporting it shouldn't be cause for much attack.... HOWEVER, making unjustified leaps of logic to explain WHY the result occurred, or what it means for other things - isn't smart (See the parallels with Kunchur here ?!?!) He heard something when recording at differing rates. The fact that his converter (may have) produced a different and audible result when resampling the output into different rates, isn't controversial at all. It shouldn't be discounted that this result could be caused by confirmation bias, and/or placebo This result shouldn't be used to "demonstrate" the general "unsuitability" of low rates (as a simple result doesn't infer causality without justification) ... but aside from that? What? There's plenty of things about how his ADC or DAC work with different rates, which could cause the difference.
eltech Posted July 6, 2016 Posted July 6, 2016 I understand. You got a better result with 192khz. You can't say why though. There may be something about how your card, or computer works with the different sampling rates, that is nothing to do with general "flaws" in lower sampling rates. Your AD converter doesn't run at 192khz. Is uses a rate of 24,000khz (24mhz) ...... when it reduces (decimation) the rate back to 48khz, or 192khz, or whatever you're asking for .... then perhaps it does something differently depending on which rate you choose. The decimation may be harming the quality when going to a lower rate ..... but this isn't becuase the "lower rate isn't capable of representing your audio properly" ..... it is because your device is not doing a good job of it. The result may be exactly the same to you (ie. the higher rate works better) ...... but it isn't correct to make sweeping statements about "high rates, yes or no". Correct, I can only report what I observe. Am I correct in thinking that you can't say why either? I agree with you that there are reasons for why things sound as they do. I am not making sweeping statements, I am simply reporting my observation, and proposing a possible reason for it. I was not sugesting my proposal is conclusive. Are you suggesting your point is conclusive? I am aproaching it from personal observation not just theory. I have many DACs from vintage R2R through to modern delta sigma 192khz capable and I have played back these files on all of them and the same observation is made - that higher sample rates make the clicks recorded from a vinyl record sound more like what is on the record. I am very familiar with the fact that sample rates affect the sound and presentation of the music played back on diferent DACs. If everything was a fact, and conclusive we wouldnt be here trying to discuss it. I asked you earlier what is your take on high rez audio. I've allready said I think there is merrit to it. Are you interested in anyone elses observations or have you made your mind up already?
davewantsmoore Posted July 6, 2016 Posted July 6, 2016 Am I correct in thinking that you can't say why either? In short yes (I can't) ... not without investigating what is happening with your converters, to actually determine the answer. ... but given that your ADC and DAC both operate slightly differently depending on which rate they are working with, then there is lots of scope for what there are differences. It's like if I got a petrol car and a diesel car from the same manufacturer / model (ie. seemingly identical except the engine) .... and I ran them around the racetrack. The petrol was always faster. Why? Well obviously we don't really know why. It's probably pretty likely to be something to do with petrol vs diesel.... but what exactly? We don't know. It would be unjustified to make statements that say... petrol cars will always be faster because diesel is flawed and doesn't work properly. In fact.... it could be that diesel is "superior" .... it is just that this specific car does something really silly with it's diesel engine ..... and so it the design of the engine, and not the liquid/gas we are combusting in it, which is at fault. It could be that ... you could make cars go equally fast on petrol or diesel fuel, if you actually designed that cars to work well (and neither diesel or petrol really are "superior") ..... and it could be that it's just "easier" to make a petrol car good.
Recommended Posts