Then I am disappointed that you chose to ignore some of my technical commentary about measurements. The one 'like' I got for one technical post was from a PhD EE who did his dissertation in chaotic sigma-delta modulators, and holds around 35 patents. Apparently he understood what I was talking about, even if you didn't see the relevance.
https://www.diyaudio.com/community/threads/the-black-hole.349926/post-7306980
https://backend.orbit.dtu.dk/ws/portalfiles/portal/5274022/Binder1.pdf
https://www.aes.org/events/141/presenters/?ID=5036
Last edited:
Mark, I ignored nothing, I just don't run a continuing commentary on everything said here, and thank you for those links, I have been making my way through them as I have time. I have only been replying to what I feel was an odd and unhelpful way of reporting what should have been a technical description of issues in your initial post on DACs.
Also, regarding a post you made a few back: No need to pretend what is heard doesn't exist; we just work on figuring our how to measure the phenomenon perceived. Calling objective quantification a "measurement story" implies the measurements are ficticious or on equal footing with subjective opinions and they are not. The very reason man has developed the scientific method was to subtract POV, bias and unrelated factors.
We have all discussed these things to death, let's head back into technical issues.
Cheers!
Howie
Also, regarding a post you made a few back: No need to pretend what is heard doesn't exist; we just work on figuring our how to measure the phenomenon perceived. Calling objective quantification a "measurement story" implies the measurements are ficticious or on equal footing with subjective opinions and they are not. The very reason man has developed the scientific method was to subtract POV, bias and unrelated factors.
We have all discussed these things to death, let's head back into technical issues.
Cheers!
Howie
Howie,
When I look into measurements, of course everything that is real is physical, and everything that is physical is in principle measurable. However there are multiple problems. Some measurement problems are easy to deal with, some are harder, some much harder, and in some cases we don't exactly how to measure certain physical things today to the accuracy needed to more forward with better understanding. In some cases we don't yet have the necessary underlying mathematical theory for some of the kinds of measurements we need to make in order to better understand audio and human perception. That's the truth.
I can go into more specific detail if you want.
So, to me some of what you may see as the good old days of measurements around here were not always actually very scientific, even if there was an appearance of science. Jakob2 correctly pointed 'appearance of science, without actual science' problem as well.
Even today, we have people who claim their views are supported by science when they don't even understand the science, don't understand what science itself actually is, don't understand the philosophy of science, and so on. And often you can't educate them. Some are so sure they are right, they ignore any explanation. Its like they are saying, "I don't need to consider your so-called evidence. I already know it cant be right, because I already know its a scientific fact that the Earth is flat!" Except they don't say, "the Earth is flat, instead they say something else that is clearly not correct, and or that is illogical, and or self-inconsistent, and or in conflict with well established existing science that they don't know about or don't understand, etc."
When I try to deal with it sometimes I wonder if Syn08 didn't have point after all (about the need to keep things simple for amateur diy'ers). But then again, we will never make any progress around here if we don't try to confront what's wrong, try to better educate people, etc.
Mark
When I look into measurements, of course everything that is real is physical, and everything that is physical is in principle measurable. However there are multiple problems. Some measurement problems are easy to deal with, some are harder, some much harder, and in some cases we don't exactly how to measure certain physical things today to the accuracy needed to more forward with better understanding. In some cases we don't yet have the necessary underlying mathematical theory for some of the kinds of measurements we need to make in order to better understand audio and human perception. That's the truth.
I can go into more specific detail if you want.
So, to me some of what you may see as the good old days of measurements around here were not always actually very scientific, even if there was an appearance of science. Jakob2 correctly pointed 'appearance of science, without actual science' problem as well.
Even today, we have people who claim their views are supported by science when they don't even understand the science, don't understand what science itself actually is, don't understand the philosophy of science, and so on. And often you can't educate them. Some are so sure they are right, they ignore any explanation. Its like they are saying, "I don't need to consider your so-called evidence. I already know it cant be right, because I already know its a scientific fact that the Earth is flat!" Except they don't say, "the Earth is flat, instead they say something else that is clearly not correct, and or that is illogical, and or self-inconsistent, and or in conflict with well established existing science that they don't know about or don't understand, etc."
When I try to deal with it sometimes I wonder if Syn08 didn't have point after all (about the need to keep things simple for amateur diy'ers). But then again, we will never make any progress around here if we don't try to confront what's wrong, try to better educate people, etc.
Mark
Last edited:
@Markw4 if this is your attempt to better educate people it's failing really badly. You've jumped onto every hot new band wagon as the best thing ever which really makes me wonder if you are the audio equivalent of Mr Toad*. The fact that you pull explainations for what you hear/think you hear out of thin air doesn't help. Then adding statements like 'Will destroy your dacs' which you would expect from someone a quarter of your age knocks your credibility.
I should also note that a 'like' neither means the liker agreed with all of the post or that you are on the right path, just that something you said was liked...
But I would like to pick at 'Localization in a stereo sound field can also vary with frequency and or be spread out laterally due to clock jitter or other effects'. To get a sound from the right of the right speaker or the left of the left speaker either requires an acoustically poor room or out of phase signals. Neither of these are likely to change between 2 competent DACs so what are these other effects and what magnitude do they have to be to give the reported increases?
*Apologies for quoting a british book (wind in the willows) which may not universally read, but my kids know about clifford the big red dog 😛
I should also note that a 'like' neither means the liker agreed with all of the post or that you are on the right path, just that something you said was liked...
But I would like to pick at 'Localization in a stereo sound field can also vary with frequency and or be spread out laterally due to clock jitter or other effects'. To get a sound from the right of the right speaker or the left of the left speaker either requires an acoustically poor room or out of phase signals. Neither of these are likely to change between 2 competent DACs so what are these other effects and what magnitude do they have to be to give the reported increases?
*Apologies for quoting a british book (wind in the willows) which may not universally read, but my kids know about clifford the big red dog 😛
Hi Bill,
Okay, I did take some literary license using the word 'destroy.' That was intended to emphasize how much things have changed over the past several years. Maybe its that dac years are kind of like dog years, 7 years may be close to a lifetime for a dac. The technology seems to keep moving along pretty fast in terms of perceptual SQ, at least relative to some other types of audio devices.
Other than that, I have been trying to figure out if there is a way out of this measurement dead end, where what we know how to do really well is measure stuff like PSS HD/IMD, SINDAD, SNR, etc. We don't have an equally good way to measure other things, as I think Hans recognizes for at least some audio devices/components. I will continue keep looking at new ideas, discussing them, getting feedback, and revising my views over time, including looking at various developments and or new/existing proposals until the dead we are stuck in starts to open up in a useful way.
For one thing, I would like to measure dynamic response of a system instead of PSS. IOW, I want to modulate signal levels. You know the problem with that, you get a spray of frequencies in an FFT to the point that the results become completely non-intuitive to a human. In the meantime we just pretend dynamics is not a problem. BS, I say. It is a problem, IMHO.
For another thing, I want to look at phase relationships between stereo channels under dynamic conditions. Why? Because I'm not satisfied with pretending its not a problem.
And it gets more complicated when sigma-delta modulators are involved. Again, we just pretend its not a problem.
I can't think of a way to politely express the extent to which its all big pile of wishful thinking measurement BS that we have now. I don't believe in it anymore, except as a gross approximation. Its a fantasy created by engineers, much like concept of ground is described as a 'fantasy created by engineers' in an IEEE sponsored book on grounding. More on that when we get around to talking about actual science verses pseudoscience.
Regarding creating wide soundstage that is wider than the speakers, I will note the following research interest:
19 - Siegfried Linkwitz, Which loudspeaker parameters are important to create the illusion of a live performance in the living room?, 113th AES Convention, Los Angeles, 2002, October 5-8, Preprint 5637, Abstract. Overhead slides of the presentation.
Please consider a live recording in a symphony hall. By your reasoning then a whole concert hall illusion can only occur in miniature between speakers or even smaller inside your head with cans. That's not a greatly convincing illusion either way.
Moreover, you seem to have omitted a common way of stretching out the stereo illusion, not that its a way I use: Shuffling.
For one example: https://www.waves.com/plugins/s1-stereo-imager
There is yet another way to produce a convincing stereo sound field that is wider than the speakers. I will wait until you get here to demonstrate with some more or less figure-8, yet curved ESL panels that are strategically located in the room. One important thing to notice is that resulting sound field is quite revealing of the quality of the recording and the reproduction electronics. Not playing games here, I really think its something that has be heard if it is not to be misunderstood. The short answer though is that the right speakers in right room can disappear as Linkwitz wanted, or at the least they can come a lot closer to that ideal than most people's systems actually do.
Okay, I did take some literary license using the word 'destroy.' That was intended to emphasize how much things have changed over the past several years. Maybe its that dac years are kind of like dog years, 7 years may be close to a lifetime for a dac. The technology seems to keep moving along pretty fast in terms of perceptual SQ, at least relative to some other types of audio devices.
Other than that, I have been trying to figure out if there is a way out of this measurement dead end, where what we know how to do really well is measure stuff like PSS HD/IMD, SINDAD, SNR, etc. We don't have an equally good way to measure other things, as I think Hans recognizes for at least some audio devices/components. I will continue keep looking at new ideas, discussing them, getting feedback, and revising my views over time, including looking at various developments and or new/existing proposals until the dead we are stuck in starts to open up in a useful way.
For one thing, I would like to measure dynamic response of a system instead of PSS. IOW, I want to modulate signal levels. You know the problem with that, you get a spray of frequencies in an FFT to the point that the results become completely non-intuitive to a human. In the meantime we just pretend dynamics is not a problem. BS, I say. It is a problem, IMHO.
For another thing, I want to look at phase relationships between stereo channels under dynamic conditions. Why? Because I'm not satisfied with pretending its not a problem.
And it gets more complicated when sigma-delta modulators are involved. Again, we just pretend its not a problem.
I can't think of a way to politely express the extent to which its all big pile of wishful thinking measurement BS that we have now. I don't believe in it anymore, except as a gross approximation. Its a fantasy created by engineers, much like concept of ground is described as a 'fantasy created by engineers' in an IEEE sponsored book on grounding. More on that when we get around to talking about actual science verses pseudoscience.
Regarding creating wide soundstage that is wider than the speakers, I will note the following research interest:
19 - Siegfried Linkwitz, Which loudspeaker parameters are important to create the illusion of a live performance in the living room?, 113th AES Convention, Los Angeles, 2002, October 5-8, Preprint 5637, Abstract. Overhead slides of the presentation.
Please consider a live recording in a symphony hall. By your reasoning then a whole concert hall illusion can only occur in miniature between speakers or even smaller inside your head with cans. That's not a greatly convincing illusion either way.
Moreover, you seem to have omitted a common way of stretching out the stereo illusion, not that its a way I use: Shuffling.
For one example: https://www.waves.com/plugins/s1-stereo-imager
There is yet another way to produce a convincing stereo sound field that is wider than the speakers. I will wait until you get here to demonstrate with some more or less figure-8, yet curved ESL panels that are strategically located in the room. One important thing to notice is that resulting sound field is quite revealing of the quality of the recording and the reproduction electronics. Not playing games here, I really think its something that has be heard if it is not to be misunderstood. The short answer though is that the right speakers in right room can disappear as Linkwitz wanted, or at the least they can come a lot closer to that ideal than most people's systems actually do.
Last edited:
I can accept sighted evaluation of a hammer or baseball bat destroying a DAC as valid technically. 🤣
Hey Mark, I totally agree...in my own experience trying to quantify advantages of different forms of dither a single number that people always seem to want to give is totally insufficient. I was told brand X ADCs dither gave 10 dB additional dynamic range, another 14 dB, a different one maybe 16 dB, etc... Yet listening to each did not agree with the quoted single-number spec, for reasons that are obvious to anyone with experience with ADCs. Pushing state-of-the-art in magnetic recording and analog audio for me has often consisted with digging deeper into performance that the specifications would indicate, so I am with you on that point.
I'd say in my experience, specs don't necessarily lie unless intentionally done, but they can over-simplify a complex phenomenon. To get to the DAC issue, I know the overall performance specs are given in single number format (dB of DR, s/n, etc) but those do not nearly tell the whole story. I have often decried the lack of comprehensive specification accompanying audio equipment.
I would say the best value-added we can do here is to try and put a finger on what sonic attributes are indeed different, then feed this information to those who may have both the test equipment and/or depth of understanding of DAC minutiae to explain it. You use imaging as an example. If two systems had the same phase response in the audio passband with no time offset between channels and identical frequency responses the imaging should be very similar. If it is not, then why? What about that device's characteristics vary from another?
Mr. Phelps, your job is to find out why...this tape will self-destruct in five seconds...
Cheers,
Howie
Nobody with even half a brain believes in your listening fairy tales especially knowing what a mess your current dac truly is. You are the only member here who constantly tries to convince others how great his audio equipment sounds. And the only member who constantly belittles professions.
Oh my oh my oh my ... another thread destined to be "blowtorched."
No doubt those on the team that started it are carefully considering what to do.
No doubt those on the team that started it are carefully considering what to do.
One problem is that we tend to measure one channel at a time, often using PSS test signals (e.g. fixed amplitude frequency sweeps). There is more than one potential problem I can think of given that type of method of assessment.
First of all, phase of one frequency may not be perfectly stable when multiple frequencies and changing volume levels are all going on at once, as they do in real music. There are changing capacitances in active devices and so forth. Of course we can use lots of feedback trying to help with that, but other unwanted effects may occur if we go to far with the feedback (For the record, I know about the 'F-word' paper by Putzeys. Agree in theory, but not necessarily in practice).
Another concern might be with power amps that share a common power transformer or even a common chassis where there can be some unintended coupling between different parts of the circuitry. That could be an issue even if the the possible issue mentioned in the above paragraph for one channel were well under control.
Bottom line is it would be great if we could precisely measure phase under dynamic (real music) conditions. IIUC right now we don't have a really good way of doing that. If we did, probably everybody would be doing it, don't you think?
Last edited:
Hi Hans,
Not acquainted with either Lars or Bruno personally. Haven't heard the Mola Mola Dac either, but would like to.
Regarding the Mola design team, found a post by Bruno:
https://www.audiosciencereview.com/...qui-dac-and-streamer-review.10770/post-300699
Mark
Regarding some possibly frustrated wording on my part about existing measurements, maybe the underlying concerns have been better expressed by others.
Here is one such effort:
"Time-frequency techniques constitutes a major improvement in signal analysis, namely at the field of biomedical signals in which the interdisciplinary nature of the proposed questions implies the development of new strategies to answer to specific problems. Time-frequency analysis using Wavelets, Wigner-Ville transform and more recently the Hilbert-Huang Transform (HHT) constitutes the core of these tools with applications in biomedical signals in last years. The non-linearity and non-stationarity nature of these signals puts HHT as a powerful tool to process signals with those properties, avoiding artefacts related to the use of linear and stationary assumptions. Classical spectral analysis using Fourier Transform still the most commonly used method when one wants to measure the global power-frequency distribution (power spectrum) of a given signal. In all areas of knowledge, Fourier-based analysis of time-series data faces constraining limitations. In biomedical signals, the critical constraining factors are the shortness of total data span, the non-stationarity of the data and the nonlinearity of the underlying physiological process. Methods using Short Time Fourier Transform (STFT) are able to extract the spectral information by defining short time windows and locally computing the Fourier transform, thereby coping with non-stationary phenomena. The frequency resolution is inversely proportional to the window length, and changes in time resolution (window length) compromise the frequency resolution. Even with optimized joint time-frequency localization, the trade-off between time and frequency resolution is unavoidable. In spite of these limitations, classical Fourier spectral analysis is still widely used to process biomedical data, for lack of alternatives. The uncritical use of Fourier spectral analysis and the careless adoption of the stationary and linear assumptions may give misleading results."
https://cdn.intechopen.com/pdfs/129...uang_transform_in_biomedical_applications.pdf
IOW, we have problems and difficulties all across science and engineering because we only have the tools we have. A problem happens if we are not careful not to let ourselves be misled, is the main point. If we look at a number such as -130dB, yes, its an impressive number, but IMHO its important to think about what the number is actually a measurement of, what isn't being measured, and not let ourselves get in the habit if thinking we measured everything that matters, or necessarily measured the overwhelmingly most important thing. Maybe we did, or maybe not. Maybe when one measurement gets low enough then it doesn't matter as much. Something more or less like that.
Here is one such effort:
"Time-frequency techniques constitutes a major improvement in signal analysis, namely at the field of biomedical signals in which the interdisciplinary nature of the proposed questions implies the development of new strategies to answer to specific problems. Time-frequency analysis using Wavelets, Wigner-Ville transform and more recently the Hilbert-Huang Transform (HHT) constitutes the core of these tools with applications in biomedical signals in last years. The non-linearity and non-stationarity nature of these signals puts HHT as a powerful tool to process signals with those properties, avoiding artefacts related to the use of linear and stationary assumptions. Classical spectral analysis using Fourier Transform still the most commonly used method when one wants to measure the global power-frequency distribution (power spectrum) of a given signal. In all areas of knowledge, Fourier-based analysis of time-series data faces constraining limitations. In biomedical signals, the critical constraining factors are the shortness of total data span, the non-stationarity of the data and the nonlinearity of the underlying physiological process. Methods using Short Time Fourier Transform (STFT) are able to extract the spectral information by defining short time windows and locally computing the Fourier transform, thereby coping with non-stationary phenomena. The frequency resolution is inversely proportional to the window length, and changes in time resolution (window length) compromise the frequency resolution. Even with optimized joint time-frequency localization, the trade-off between time and frequency resolution is unavoidable. In spite of these limitations, classical Fourier spectral analysis is still widely used to process biomedical data, for lack of alternatives. The uncritical use of Fourier spectral analysis and the careless adoption of the stationary and linear assumptions may give misleading results."
https://cdn.intechopen.com/pdfs/129...uang_transform_in_biomedical_applications.pdf
IOW, we have problems and difficulties all across science and engineering because we only have the tools we have. A problem happens if we are not careful not to let ourselves be misled, is the main point. If we look at a number such as -130dB, yes, its an impressive number, but IMHO its important to think about what the number is actually a measurement of, what isn't being measured, and not let ourselves get in the habit if thinking we measured everything that matters, or necessarily measured the overwhelmingly most important thing. Maybe we did, or maybe not. Maybe when one measurement gets low enough then it doesn't matter as much. Something more or less like that.
It's not my reasoning. You said CDs are lossy. They are not. Any media may have had lossy processing prior to recording - do you understand that difference?
Sure, I just thought we were talking about audio reproduction using audio CDs. The overall result is lossy. Do you disagree on that?
I have a Aqua La Scala DAC in the house right now and with it's ladder and tubes it sound very smooth. I think it is the best sounding DAC in its price range.
Mark, by this metric all record media are lossy, including 32 bit 348k...it is just a matter of diminishing error. Agreed?
Mark- There is a technique used for validating data links. It uses pseudo-random data and compares the output to the input. On a simple data link you can get errors in the PPT range. The concept could be adapted to audio links. The pseudo random noise should test every possible change in the audio. With the corrections for delay and frequency/phase at the analog level anything remaining as a difference reflects degradation in the audio link. A short variant would be to apply a digitally generated pseudo-random noise to two different DACs, adjust for delay and any obvious phase amplitude response and measure differentially.
However at the other end of the "spectrum" the lossy audio developers have figured out a lot about what our hearing is sensitive to and what it is not sensitive to. Because lossy is a perjorative term here we are ignoring a body of probably useful information.
However at the other end of the "spectrum" the lossy audio developers have figured out a lot about what our hearing is sensitive to and what it is not sensitive to. Because lossy is a perjorative term here we are ignoring a body of probably useful information.