 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
97.118.124.106
In Reply to: RE: I am amazed posted by Say on June 15, 2013 at 13:31:01
...is the fact that people that don't hear any differences waste their time coming to a place like this where people come to discuss the differences we hear.
I guess he just likes to hear himself talk. I have to admit that is the first time I have seen the word "ideation" used outside of a refereed technical journal. Impressive. (Yawn.)
Follow Ups:
> And even more amazing...
> ...is the fact that people that don't hear any differences waste their
> time coming to a place like this where people come to discuss
> the differences we hear.Charles, A few years ago, you and Gordon were eager to talk about technical details and measurements of async vs. adaptive mode vs. SPFIF.
You may now feel that the only topics that should be allowed here are swapping reports of differences. Fortunately, neither you nor any of the other people posting on this forum gets to dictate what can be discussed here.
Bill
my blog: http://carsmusicandnature.blogspot.com/
Edits: 06/15/13
Hi Bill,
I have to say that I have something of a love/hate relationship with forums, and this forum in particular.
I find that the general level of experience, knowledge, and depth of background is quite a bit higher here than just about any other forum. You are a perfect example of this, and I feel that I have probably learned more from you than the converse. For this reason I love this forum (Audio Asylum in total, not just the Computer Audio sub-forum).
It is great when people from around the world can freely exchange their ideas, whether or not they disagree.
But the thing that I hate about all forums (and the Audio Asylum in particular) is that the same, tired, arguments that can never be settled over the internet constantly are repeated over and over and over and over and over and over and over and over and...
The reason that it is worse here than any other forum is the unique way in which the threads are presented. In the default view, a thread only stays visible for a few weeks (at most) before dropping off the bottom of the first page. And like they say in bicycle racing, "Out of sight, out of mind." The result is that at least once a month the very same discussion we all just went through is raised again by somebody who missed it the first time.
Even in the alternative views that mimic most other forum sites and bring the thread with the most recent comment to the top (rather than the most recent starting date) suffer from the same problem. Which is why I posted "Yawn". How many times to we need to have the discussion that things that are bit perfect do or don't sound the same?
This is question that can ONLY be resolved by personal experience. Either a person has had an audio experience that cannot be explained by our current understanding of electronics and/or acoustics, or they have not. I think that all of us with a scientific background start out on the "skeptical" side. And then at some point we either have that pivotal experience or we don't.
I consider myself fortunate that I had it at a very young age (around 20). But after attending college and graduating in physics, I felt that I knew more than I really did. And so I had to overcome my prejudicial attitude again. "There's not point to make a listening comparison to THAT, as it couldn't possibly make any difference whatsoever. Simple physics tells us so!".
I metaphorically had to get hit over the head with a sonic 2 x 4 to recognize that REGARDLESS of my prejudices, that such-and-such a change actually DID make an audible sonic difference. But my educational background has been helpful in different ways also. Yet it has also helped me to design experiments where only ONE variable is changed at a time. (I am constantly surprised by the number of so-called "professional" reviewers who haven't figured this out yet....)
Gordon and I were happy to discuss the advantages of asynchronous USB when it was a new discovery that nobody knew about. But now that it has been adopted by almost every reputable manufacturer, and there is an extensive white paper on our website explaining it in detail, there doesn't seem to be much point in continuing those particular discussions.
I suppose my point is that I am interested in discussing new ideas, new observations, and new knowledge and NOT so interested in discussing the same old tired point repeatedly, especially when there is no conclusion.
I was recently sucked into a discussion over at Computer Audiophile because someone had posted misinformation regarding our upgrade to our USB DACs. It has been a typical thread where the topic under discussion has changed several times. However, there was quite a long time where it became a discussion of the merits of PCM versus DSD.
I participated in that because when developing our A/D converter (along with our experiments with digital filters on the playback side, I felt that I was able to offer new insights into the merits of each. Naturally there was the usual group of people who didn't want to have a discussion of ideas, but rather an affirmation of their chosen beliefs (much like religious fanatics), but there was also a fair number of people who greatly benefited from the discussion. (They posted so explicitly, and the rule of thumb is that for every person who takes the time to do so, there are at least ten others who feel the same way, but don't take the time to post.)
If you would like to read it, that discussion is at the link below. You may find it worth reading, although many parts of it are not. I just grow weary of the same unanswerable questions being raised time and again. I certainly don't think that I (or anybody else) should limit that range of topics that can be discussed here (or anywhere else). However, it WOULD be nice if people would use the search functions on the site to see if the topic has already been covered before posting. That's all. Good to hear from you.
Best regards,
Charles Hansen
> You are a perfect example of this, and I feel that I have probably
> learned more from you than the converse.
Thanks. I enjoyed our conversations and benefitted from them.
> This is question that can ONLY be resolved by personal experience.
> Either a person has had an audio experience that cannot be explained
> by our current understanding of electronics and/or acoustics,
> or they have not.
I feel that a healthy scepticism is a good survival mechanism. When it hardens into dogma, that is not good. I come across things that I don't understand quite often. I try to note that lack of understanding and act accordingly. Some things I just don't care enough about to investigate.
However, what I see in audiophile discussions is that unlikely audio experiences are rarely investigated in a thorough way to establish a cause and effect relationship.
You talked about experimental technique in your post. We could use more good design and good interpretation.
> Gordon and I were happy to discuss the advantages of asynchronous USB
> when it was a new discovery that nobody knew about.
I think you were a big factor in generating interest in async USB and that's good.
> I suppose my point is that I am interested in discussing new ideas,
> new observations, and new knowledge...
Well, computer audio isn't new anymore. I've had my muic collection on a computer for about 7 years now. However, there are some things left to talk about.
I'll email you about a topic that you and I talked about some time ago. It isn't a question of new knowledge but about a way to make progress on a useful architecture for computer audio.
> If you would like to read it, that discussion is at the link below.
I read a few pages of the CA thread. Lots of people jumping to conclusions and reasoning from those faulty conclusions. That gets tiresome.
One post quoted from something you wrote in March:
"Addition of an AC line powered supply for the USB circuitry. This provides for uniformly superior performance, regardless of the quality of the USB Vbus power supplied by the computer."
It sounds as though there is a story there. I believe that the USB receiver side was powered from USB in the original QB-9.
The quote reminded me of a discussion of problems with USB power for a Raspberry Pi computer module. Link below. Laptops and headless laptops (Mac Mini) are popular for use as music servers but I wonder about the quality of USB power they provide. Conserving power is a very powerful design objective for laptop makers and that might lead to compromises that aren't good for music server use.
You also said that the ESS chip is a huge step up from the Burr Brown chip that you used in the original QB-9. More detail characterizing the difference in performance would be welcome.
Bill
my blog: http://carsmusicandnature.blogspot.com/
Hi Bill,
Thanks for the thoughtful post.
> > I feel that a healthy scepticism is a good survival mechanism. When it hardens into dogma, that is not good. I come across things that I don't understand quite often. I try to note that lack of understanding and act accordingly. Some things I just don't care enough about to investigate. < <
If by "good survival mechanism" that you mean you don't accept all manufacturer's claims blindly, as otherwise you would:
a) Go bankrupt in short order purchasing every "new" wonderful invention.
b) Own at least two or three products in every category because the various manufacturers invariably make conflicting claims.
Then I agree completely. On the other hand, there are hundreds of end users her that post of hearing differences between various bit-perfect players. They have no commercial agenda. There tends to be a general agreement on what the various player sound like, even if that doesn't lead to a universal preference for one over the other. To me that strongly suggest that there *are* audible difference and he lack of agreement on which one is "best" merely boils down to which one is "best" for that particular listener's tastes or characteristics of other components in his system.
From my point of view, the "hardening into dogma" comes from people like Archimago. They have already decided that all bit-perfect players sound the same, and that is exactly what they experience. I don't if that is due to an expectation bias, a poorly set-up system with low resolution, poor testing methods, or whatever. But when I have performed carefully-controlled experiments (often blind - but perhaps not double blind) and clearly hear those difference for myself, it is very difficult for me to say that the people who *do* hear a difference are the ones whose experiences have "hardened into dogma".
Every time I have tried to investigate the mechanisms, I make zero progress and end up wasting my time. For me it is enough to know that a player that converts the integer data on a CD to floating point and back again will degrade the sound quality. I don't need to know WHY. It's the same with gravity. It is enough for me to know that if I let go of an otherwise unsupported object that it will fall -- every time. I don't care that nobody knows how gravity works or what the mechanism is. I'm not about to sit down and try and figure it out. There are minds far superior to mine who have spent entire lifetimes on such problems without success.
So I treat known facts in audio the same way I treat gravity. RFI degrades the sound of audio components. Some are more sensitive to this problem than others. But I know that if a DVD player has a "Display Off" button that turning it off will improve the sound - every time. I know that increasing the RFI levels by using Wi-Fi wil degrade the sound -- every time.
Now I rarely bother to turn the DVD'S display off, as it is not worth the inconvenience. So I can understand that people will use Wi-Fi, because they will accept the degradation for the gain in convenience. Look at all the cell phone users who saw Ted Kennedy die of a very rare brain tumor that either he or his oncologist (or both) blamed on very heavy cell phone usage -- yet they continue to use cell phones for "the convenience". If you ask these people, the will typically give the same reply as the smoker does, "Something's going to kill you sooner or later anyway. May as well live my life the way I want to and deal with the consequences if and when they come."
> > However, what I see in audiophile discussions is that unlikely audio experiences are rarely investigated in a thorough way to establish a cause and effect relationship. You talked about experimental technique in your post. We could use more good design and good interpretation. < <
I have talked about this many times. Most people pay no attention. I have sound certain rules that must be followed:
1) When making comparison, be fanatical about changing just one variable at a time. So if we are comparing two source components. we will set the two side-by-side on identical racks. We will make sure that they are both fully broken in by playing them on "repeat" for at least a week. We will listen to at least three songs on each source before switching to the other source -- quick A/B comparisons are useless. Trying to compare more than two components at a time is useless. We use the same interconnect cables, connected to the same input on the preamplifier and move the cables from one component to the other. After each movement of the cables, we play a 1 minute glide gone (Track One of the Ayre IBE disc) to "settle in" the cables after they have been disturbed by moving them before listening to the music.
The single most important thing when doing these tests is an intimate familiarity with the musical selections. Without this, the tests are useless. The next most important factor is familiarity with the system. The third most important thing is the manner in which one listens to the music.
It is useless to try and remember every single detail about every single sound made by every single instrument and voice in the song. Humans are like crows. We can only remember up to seven objects. That is why telephone numbers are 7 digits long. Instead one must listen to the song for the sheer enjoyment of doing so. It is the entire reason that humans listen to music in the first place. Then when the song is played through the other source, listen the same way.
One of the sources (or whatever change you have made the the system) will be more engaging than the other. The better component (or whatever change you have made to the system) will draw you in and make it difficult to turn the music off. With the lesser component (or other change to the system) you will find you mind wandering. You will begin thinking about unrelated things such as what time it it, how hungry you are, what bills are coming due -- anything but the music itself.
One might thing that this is because you are bored having just heard the song if it is the second source (or change to the system) that sounds worse. It is easy to disprove this, because one can switch back to the first configuration and fine yourself just as engaged as the first time.
Then after you have determined which configuration sound more engaging, think back and remember what it was about the sound that was different. Was it that the bass was anemic and hard to follow the lines? Was it that the upper midrange was edgy and grated on one's ears? Was it that a lack of detail homogenized the individual musicians? Was it that some sort of subtle timing error made it sound as if the musicians had not been playing together in the same room at the same time?
For the end user you stop there as there is not much else to do. For the circuit designer, your work has just begun. as now one need to dig down to a deeper level and find out WHY one capacitor sounds better than another. (Or one could just limit oneself to using brands and models of capacitors that have been approved by listening tests.)
> > Well, computer audio isn't new anymore. I've had my muic collection on a computer for about 7 years now. However, there are some things left to talk about. < <
I agree completely. However I don't think that the audibility of differences in various software music players that are all "bit perfect" is one of them By now one has either had that experience or they haven't But when someone comes in and says "I don't even have to listen to know that they are all the same, because they are all bit perfect -- ergo, they MUST sound the same", I don't find that a topic worthy of discussion.
The trajectory of such a thread is entirely predictable. The alleged "scientist" is anything BUT. A true scientist makes observations and then attempt to explain them. Archimago comes in with preformed conclusions at worst, and faulty test methodology at best, and doggedly disagrees with all the people who agree that gravity exists (even though we can't explain it) because every time we let go of something, it falls.
And all the people that have had something fall say, "Why don't you try dropping something and see what happens? And the "scientist" will say "I don't have to as I already know what will happen." Or he will say, "I did and it still didn't fall." And then the other people will say, "Try it again, but this time don't glue the object to your hand." And on and on it goes....
> > It sounds as though there is a story there. I believe that the USB receiver side was powered from USB in the original QB-9. < <
Yes, I have already explained it but perhaps not in this forum, I don't recall. We originally used the computer USB Vbus power to supply our USB circuitry. This was to save money. We felt that we didn't need to do anything else, as the USB circuitry is galvanically isolated from the audio circuitry (and the rest of the user's audio system) by a bank of high-speed opto-isolators.
Over the years we have had a few people tell us that further isolation improved the sound. We don't like to make running changes in our equipment. We don't want people to say "MY XYS-11 sounds better than your as it has the new higher speed furbulators in it!" Instead, we save up many small changes for an update to the product.
Adding DoP was a very visible update so we took this opportunity to try many different things. One of them was powering the USB bus from another source. One customer had said that when powered with DuraCell NiMH batteries that it sound incredibly better. but with Eveready NiMH it made no difference. So we duplicated his experiment as best we could (he stopped corresponding with us at one point). In our system, both batteries sounded the same and slightly worse that the Vbus from a MacBook Pro.
Then we tried a simple AC supply with a 3-pin high-feedback regulator. That sounded very slightly better. So we tried an AC supply with a discrete zero feedback regulator. It sounded identical to the 3-pin regulator (not what I was expecting). We conferred with Gordon Rankin and he said that Mac laptops have just about the cleanest Vbus power of any computer. We couldn't easily test with all computers. All the ones we have at the factory are either Macs or Lenovo ThinkPads, which also use very high quality parts. But it wasn't that expensive to add the additional isolated Vbus transformer and power supply. Then even if a customer had a very poor power supply, they would always achieve the same level of sound quality. So that became part of the upgrade.
> > You also said that the ESS chip is a huge step up from the Burr Brown chip that you used in the original QB-9. More detail characterizing the difference in performance would be welcome. < <
Actually I said it was *another* step up, not a "huge" step up.
We were preparing the upgraded DX-5 for the recent CES. As usual things take longer than expected and we had a hard deadline when the last person was leaving via plane. All three of our engineers stayed up all night working on the prototype, getting everything ready. The DoP was just a firmware change and had been working for weeks. We made some changes in the analog circuitry that we had learned from building the AX-5 integrated amplifier. We made some changes in the interface between the DAC and the analog output stage that we had learned from building the QA-9 A/D converter. Both of these together made a "huge" difference by themselves.
Then we tried the ESS DAC chip. This was a much bigger change as it required adding a daughter-board due to the different package size, pinout, and voltage requirements. Additionally it required a change in the clock frequency for various reasons.
When we changed to the ESS DAC a few of the register setting were not correct and we weren't getting output. We only had a few hours left before the plane was leaving for Las Vegas. So we thought "Let's just try it with the stock ESS digital filter and see how it sounds."
In a word, the answer was "Frustrating." One could easily hear improvements in coherency and musical naturalness from the improved modulator (see link below). Yet, the drawbacks of the stock digital filter were painfully obvious to us. Luckily with some help from Dustin Forman of ESS, designer of that chip, we were able to get everything working. And then we had the best of both worlds. But honestly, if we had to send it to the show with the choice of the ESS DAC and its stock digital filter versus the Burr-Brown DAC with our new upgrades and our custom digital filter, I would have gone with the Burr-Brown DAC. In other words, in my opinion our custom digital filter is more important for sound quality that the improved modulator design that ESS has developed. YMMV.
But given that we can purchase a B-B DAC chip for ~1/10 the price of the ESS, and even when the cost of the Xilinx FPGA is included (for the digital filter) the B-B + Xilinx is still less than half of the cost of the ESS, to me the ESS only makes sense for a product that has a generous budget allowance for the part in the BOM.
Hope this helps,
Charles Hansen
Charles,
Typing is not easy for me now so I took some time to reply.
Thanks for the details on your USB power and ESS chip changes. For me, reading such information is the best part of this forum.
The skepticism I mentioned has served me well. I try to make informed choices in audio gear and keep the gear until it breaks or my needs change. That skepticism backed by some knowledge helps me detect BS.
> From my point of view, the "hardening into dogma" comes from people
> like Archimago.
I rarely see people describing views that are inadequately support by facts but that support their own beliefs as dogma.
Archimago draws broad conclusions that aren't warranted by his experiments. I'm happy to see someone posting some results of experiments other than sighed listening tests. I can take the conclusions or leave them. I note that there has been much criticism of his tools and his methods but little posting of better quality results. And I see very little criticism of sweeping conclusions based on far less experimental evidence as long as the conclusions support the "everything matters" view.
> For me it is enough to know that a player that converts the integer data
> on a CD to floating point and back again will degrade the sound quality.
I think that you and I had some conversations about this topic in the past.
In a very minimal player, there is no need to convert the audio data back and forth. I'd be happy with a player that just passed the integer data directly to the driver. If I were writing my own player s/w, I'd keep the data path simple without conversations to and from float.
That said, if you use 64 bit float, the conversions may not alter the audio data that goes to the DAC. It would be straightforward to capture the bit stream before D/A conversion and do comparisons.
If you want features like DSP, room correction or digital crossovers for multi-way speakers, floating point arithmetic is probably a better choice than fixed point arithmetic.
I don't know what players you have compared to decide that players that convert to float and back again are inferior.
> I don't need to know WHY.
I disagee strongly with that point of view. Pursuing an observation until you understand the underlying cause and effect relationship is the basis for progress in science and technology.
Understanding why all bit perfect players don't sound alike might lead you to improve your DAC so that it sounds better with any of those players.
Bill
my blog: http://carsmusicandnature.blogspot.com/
> > I don't know what players you have compared to decide that players that convert to float and back again are inferior. < <
It was a prototype version of what became Decibel.
> > I disagee strongly with that point of view. Pursuing an observation until you understand the underlying cause and effect relationship is the basis for progress in science and technology. < <
Yes.
> > Understanding why all bit perfect players don't sound alike might lead you to improve your DAC so that it sounds better with any of those players. < <
Running a business is much different than applying for grants to do research at an institution of higher learning. The constraints don't allow for the same luxuries of time, nor the same mindset. I have done some research (as an assistant) and it was great fun. We were looking at the isotopic composition of extra-solar-system meteorite samples. This gave us clues about the formation of the solar system. Very interesting stuff, but does not pay the bills....
Cheers,
Charlie
> It was a prototype version of what became Decibel.
Oh, in the Mac environment.
The difference you heard may have been caused by the conversion to/from float or may have been caused by some other difference in the software path the audio data followed.
It isn't trivial to understand what is happening inside software you didn't write when you have neither source code nor internal documentation. Developing my understanding of Windows internals was very useful in many of the projects I did.
> Running a business is much different than applying for grants
> to do research at an institution of higher learning.
My experience doing scientific research fit right with earning a living writing computer software.
Most of the consulting projects I did involved some aspect that had not been done before. I always began by identifying the "showstopper" issues that might cause the project to fail. I did experiments to understand whether the project project could be successful and how I might proceed. That phase was quick and then the rest of the project went smoothly with low risk of failure.
Bill
my blog: http://carsmusicandnature.blogspot.com/
The conversion process between 32 bit floating point and 24 bit fixed has two different aspects that may affect the sound.
1. While the conversion from 24 bit fixed to 32 bitfloat is a one to one function, that is not the case going the other way. This creates a design decision, a.k.a. a potential problem. There are at least three strategies that may be used, and the choice may affect the resulting bits. The issue concerns what happens to the extra (low order) bits that may be present in the floating point format: (1) Truncation. (2) Rounding, (3) Dithering.
If the floating point was generated directly from the 24 bit fixed point format then Truncation and Rounding will return the original bits. One might say that for this case either one would be the optimal strategy. Unfortunately, if the floating point were generated by floating point processing (e.g. a sample rate conversion, equalization, or even a simple gain change) then truncation or rounding will introduce distortion components into the audio. This distortion can be eliminated by using a dither operation when converting back to integer. This is (usually) the best sounding strategy for these recordings. Unfortunately, introducing dither in the unneeded cases is a bad decision, as it adds unwanted noise.
Unfortunately, there is no "magic" way of knowing which type of recording is which, and hence there is no single best strategy that should be used in general. For this reason, high quality software packages allow options that the user can set. Some people may say that changing dither at 24 bit levels is irrelevant because it can not be heard. Others, say that these changes "completely trash the sound". So there is debate not just about what to do, but whether or not it even matters. The obvious solution to this quandary is to avoid the entire question by going straight from the original integer format to the DAC.
2. Although the conversion process (particularly truncation or rounding) is very quick because of hardware support, there will be additional overhead associated with doing conversion, especially considering that data will have to be repacked and hence moved. Therefore, there will probably be a fair amount of processing associated with a conversion between formats, as would be needed, for example, when sending floating point data to a DAC. (This might be less when converting 32 bit floating point data to 32 bit integer and sending that to a 32 bit integer DAC.) This processing overhead will affect electrical noise in the computer and that may leak out to the analog audio stream going to one's speakers. For those who believe that "bits are just bits" this argument is irrelevant, of course.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
I'm comfortable with 32 bit float when the original audio data is 16 bit samples. With 24 bit samples, 64 bit float seems a good idea.
1. I made a similar point in the JRiver forum thread linked below. Matt responded and Bob Katz supplied some experimental information.
2. It is natural to think of uncompressed audio as flowing from a disk file to the DAC chip with no processing and no or very few data copying steps. In practice, things often get messy and several copying operations may be needed in the minimal case. I counted 4 or 5 copy operations that might be present without integer/float conversions.
Yeah, conversion to and from float is extra work in the computer. It's just one step in a process that already involves several copy steps.
Bill
my blog: http://carsmusicandnature.blogspot.com/
For simple gain changes, etc., 64 bit float is probably not necessary. It is definitely needed for serious DSP such as sample rate conversions. Note that 32 bit IEEE floating point format actually has 25 bits of precision (In addition to the 23 + 1 bits described in the reference below, there is the sign bit.)
" The typical precision of the basic binary formats is one bit more than the width of its significand. The extra bit of precision comes from an implied (hidden) leading 1 bit. The typical floating point number will be normalized such that the most significant bit will be a one. If the leading bit is known to be one, then it need not be encoded in the interchange format."
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
> For simple gain changes, etc., 64 bit float is probably not necessary.
For simple gain changes, float may not be necessary. :-)
Bill
my blog: http://carsmusicandnature.blogspot.com/
Am i correct in concluding that you ended up dispensing with the FPGA filter mechanism because you figured out how to put your own bespoke filters into the SABRE chip filter engine?
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
The Saber is very limited in what you can do with it. There is a two section digital filter. The first section has something like 124 taps and the second section has something like 24 taps. This is just off of my very bad memory, so I could be off.
But we do all of our filters in one pass and use up to 512 taps on some of them. There is now way to do what we want in the Saber. Too bad -- it would save us some money....
I had deduced (rightly or wrongly) that the Incita and Mirus are re-purposing the filters in the SABRE chip. (This would seem possible in any event, because it is likely that the chip designers would have made the filter coefficients controllable, if only via JTAG access for diagnostic purposes.) That's why I asked the question.
Depending on the details of the two step processing it might (or might not) be possible to get similar filter responses. Whether the resulting filters would sound the same as the ones you presently use are another question.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Post a Followup:
| FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
||||||||||||||
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
