Computer Music Audio Quality
bramankp -- Tue, 12/27/2011 - 08:51
Published in TAS #219, I'm having some real problems with this article co-authored by Charles Zellig and Jay Clawson.
While most of the article is actually quite good I have a very hard time with the following sentence, "Although JRMC reported an accurate rip for all reading speeds, and are bit-for-bit identicle at all reading speeds, we are still able to detect sonic differences in the resulting files."
Really? Maybe that's not exactly what the authors meant to say but it's the equivalent of saying one can make a copy of a ripped file and somehow the exact copy sounds different. I want to trust what's being presented in the article but this is a sticking point for me. The same bits (*the same bits*) played back from a different position on the hard drive sounds different?
Someone please clear the air on this.
All content, design, and layout are Copyright © 1999 - 2011 NextScreen. All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
This is a follow-up to my recent post of 2/27/2012 re: Charles Zeilig's article, and one of his findings where he found that removing extraneous USB connections from a Windows PC significantly improved sound quality. (See comment no. 5 in my post.) I happened upon a possible answer as to why this improvement occurred. Let's call it the lightly loaded USB hub explanation. Berkeley Audio Design, a leading company in digital, advises in their Berkeley Alpha USB user guide:
(quote) "Not all USB connections on a computer are equal. USB does not have a guaranteed latency mode, so for a real time task like audio delivery, it is important to connect the Alpha USB interface to a lightly loaded hub. All USB devices and where they are connected can be seen under 'About This Mac' under the Apple Menu. Click on it, and then on 'More Info', which launches the system profiler. Click on 'USB' under Hardware to see the full list." (end quote)
BTW, The Well-Tempered Computer makes a similar observation found deep at their web site:
(quote) "Best practice is to connect the USB DAC to a USB port on your PC not shared with any other device." (end quote) (http://thewelltemperedcomputer.com/SW/Windows/Win7/USBDAC.htm) How to check Windows PC USB internal hub connections is shown here too.
Here are some other general tips in the Berkeley Audio USB user guide:
(quote) "It is good practice to keep various types of signal and power cables physically separated from
each other. This is especially important for the USB cable to the computer. It should be routed
away from the chassis of the Alpha USB and any other component or cable that is part of the
audio system. 1.5 meters is a good default length for USB, SPDIF and AES cables." (end quote)
(quote) "For optimum audio quality with a Mac, use of high resolution music server software such as Pure Music is highly recommended. The Pure Music application uses iTunes to manage files and as a user interface, but the audio data is played through its own engine, bypassing core audio in the Mac for bit-perfect output. It also has the ability to switch sample rates on the fly in full coordination with the Alpha USB and Alpha DAC, while maintaining bit accurate transparency." (end quote)
(quote) "Vista is not recommended. Windows 7 works very well as an audio server and is preferred over XP." (end quote)
I hope this info helps anyone who is tuning in their computer front end for best sound quality.
[quote] doesnt get decided soon[/quote]
it was ALREADY DECIDED BY THE PERSON THAT CODED/AUTHORED THE CODEC!! IT HAS NO QUALITY DIFFERENCE FROM WAV!! None. identical.
jesus christ. i need a tablet. the BS in here is thick.
I cannot believe that TAS would actually publish such an irresponsible article. The audiophile world has a difficult enough time as it is with credibility in the eyes of many. This series of articles, written by people who are apparently ill-informed about computers and digital audio in general, is an embarrassment.
For the record, I did try to replicate their results. I ripped three different tracks from CD to WAV using dBPoweramp on a Samsung SH-S223Q :
"Time" from Pink Floyd's Dark Side of the Moon
"Very Early" from Allan Holdsworth's None Too Soon
"The Cellist of Sarajevo, Op. 12" from Yo-Yo Ma: Solo
I did the WAV -> FLAC -> WAV conversion process (using FLAC 1.2.1b set on level 8) and compared the resulting WAV to the original and found them to be identical: MD5 showed bit-for-bit identical files, reversing phase in Soundforge showed identical waveforms/total silence, and subjective listening using Foobar2000's ABX plug showed I could perceive no audible differences (through either my Genelec 8030As or my Senn HD650s). So, I went ahead and did the repeated loop conversion of the original Allan Holdsworth WAV into FLAC (level 8) then back into WAV 1000 times (as per post above), and the resulting WAV compared to the original was likewise identical.
"So, I went ahead and did the repeated loop conversion of the original Allan Holdsworth WAV into FLAC (level 8) then back into WAV 1000 times (as per post above), and the resulting WAV compared to the original was likewise identical."
How could that be possible? I mean they can't be identical after all those conversions. Then they would sound the same ;)
Thanks for doing that. I was waiting for someone to show that proof, although I'm sure Zeilig and Clawson could have found plenty of evidence like that to question their findings by searching Hydrogen Audio, enough that they would have known better to publish it with such glee and lack of details.
You know, I kind of feel for Zeilig and Clawson. I think they may truly believe their conclusions, and their lack of basic understanding of the way computers store audio information allows them to "stand behind" those conclusions with a straight face. Yes, they should have done more of their homework on computer audio file formats -- and basic arithmetic for that matter -- prior to finding a publisher. But I blame TAS most of all for allowing such nonsense (and easily provable nonsense, at that) to be published in their magazine. At this point, it is simply FUD being flung about by people who understand this subject incompletely. Now I bet that it will become "common wisdom" among audiophiles that FLAC degrades the original signal, and there will be no arguing with them now that TAS has endorsed this demonstrably false assertion.
I would love to see an editorial in TAS from any of the countless respected authorities on digital audio who understand why this article is so flawed. I wonder if Robert Harley is brave enough to seek the truth of the matter, even if it makes these writers -- and himself by extension -- look less than authoritative about something.
It's good to give people the benefit of the doubt, but I don't buy it. Zeilig lists oncology research as his profession on LinkedIn and cites educational experience in various medical schools. I don't understand how a medical researcher--a scientist--could possibly have thought what these two published was a good idea. LOL
TAS circulation isn't that much anymore (and will probably fall more after this debacle), and even without that qualifier it doesn't appear too many audiophiles are taking this seriously.
I registered an account to chime in on this discussion. Although not an audio pro I've been doing computer based audio in various forms for upwards of 17 years now. (I can't believe I paid $1500 for a SCSI-U2W drive fast enough to serve data to burn $10 CDRs back in the day!). I've built over 30 computers and used much of the prosumer gear (and better...and also worse) out there over that time.
I actually have to applaud Messrs. Zeilig and Clawson for the amount of effort they put into their study, and the apparently rigorous approach to the parts they did understand and could control well. I also feel that there is a lot of very useful information in the 4 part series that I am excited to test and implement in various systems. The hard work on listening to various software playback systems, various other permutations of configurations, etc., could actually lead to improved sound. I say "could" because it's just food for thought, since unfortunately they didn't eliminate enough of the variables INSIDE the computer to give the study any real "scientific" credibility. But I do appreciate the effort, and the information presented in many parts of the article is far from worthless. If it had just been posted on their own website it'd be a great piece.
However, by the time I finished all 4 parts of the article, I too was wondering how TAS could present this as a scientific and rigorous study with any engineering cred. This just wasn't appropriate to a professionally edited audio magazine that people pay to read. In other words, it has me wondering if anyone at TAS actually understands computer audio in the slightest, given the green light to publish this.
Not to say, as above, that there aren't a number of possibly very useful tidbits of information, but the study as a whole just falls apart when you see their discussion of possible causes and their conclusions as to how the differences in sound quality came about (and yes, I am also a bit skepitcal about their quality scale -- but that is an attitude I hold across all audio reviews, not just this particular study, so that's just inherent in the hobby).
In effect, their final conclusions regarding permutations of process and the resulting differences in sound quality are almost primordial. Something like "when we sacrificed a calf, there was a lightning storm, and when we sacrificed a virgin, the volcano erupted." Of course we all know now, yes there is a difference between a lightning storm and a volcano eruption, but we know the calf and the virgin had nothing to do with it. I feel their understanding of what goes on inside a computer is about as enlightened as the caveman's understanding of weather. To them it's a demon black box and I believe their own expectation bias allowed some of the results they got.
The other possibility is that they didn't dig into the software configurations deeply enough. Maybe they had some DSP settings on, or at least normalize, and when they did their FLAC-WAV-FLAC (etc.) conversions, and there really was some processing that degraded the sound. Therein lies many of our problems with the study - they never even discuss whether they confirmed the files were identical.
So I agree with those who posted more of an indictment of TAS than those who indict the authors. The authors knew what they knew and tried what they tried, and it generated some interesting debates and thought. But TAS, what were you thinking?
Just my two cents. I look forward to your thoughts.
interesting debate??? What is interesting about two clueless audiophile press members attempting to rewrite the laws of physics??
interesting? not at all. It is sad and pathetic. I tell you as I tell the others. Read the books I mentioned earlier. they tell you all you need to know.
one gets to have their own opinion, but not their own facts...and it cannot be refuted that lossless codecs are IDENTICAL sonically to WAV... Computer software sounding different? Again. Please learn about DSP and digital audio. Burning speeds affecting sound? Laughable.
ALl one has to do is READ THE TEXTS!! the FACTS, irrefutable FACTS have been established for years..beginning in the 30s.
Why are audiophiles delusional? do you behave this way so that you will have something to chase?? is it painful to accept that things are as good as it gets, nothing to tweak, gear hypochondria??
Utterly pathological. It makes me want to pluck my hair out by the roots..the sheer ignorance, unbridled, festering ignorance...and the authors shirk responsibility by writing such twaddle. are they really that ignorant?? either they know better and write in spite of that to try to keep their industry afloat(keep their audiophile press jobs) or they truly are ignorant...which i doubt anyone could be that ignorant.
Umm, I was essentially agreeing with you. However I don't think you can characterize the study as entirely worthless as you seem to imply. I do agree that lossless is lossless, but that does not mean that there could be no audible difference between decompressing a FLAC file real-time and playing a WAV file. Nor does it mean that there aren't ways in which various software applications differ in their approach to rendering audio or any included processing. I agree the whole thing about the quality deteriorating through multiple WAV-FLAC-WAV conversions is silly. However I don't agree that there are not ways to optimize the sound of computer audio and that would include optimizing certain hardware and software components, and configurations thereof.
Dr. Zeilig,
I would like to come back to the FLAC-issue once more... I am curious about whether or not you have tested (direct comparison) other lossless formats like ALAC, APE etc. to see if the same type of degradation occurred.
Peter
Jesus Christ man!! THERE CAN BE NO DIFFERENCE. IT IS NOT POSSIBLE! Please for the love of common sense and education, LEARN THE PRINCIPLES OF SOUND REPRODUCTION! Nyquist/shannon theorem. Principles of digital audio!
Why do you people refuse to accept the absolute, UNREFUTABLE TRUTH when it stares you in the face???
Why are audiophiles such fools??
Maybe it is time for another tablet? Or maybe two? And perhaps some new glasses that allow you to read my prior posts?
You are barking up the wrong tree here buddy. My dogs sound way better than you. That is highly subjective, of course...
Peter
I agree that many audiophiles are fools. I also believe, based largely on my subscriptions to Stereophile and TAS that many audiophiles are from a bygone era and just simply don't get what goes on inside a computer. However, I think such an absolutist stance is really the equivalent of being a flat-earther. What NCdrawl implies is that the digital sound off a motherboard SPDIF output is necessarily the same as the sound from the SPDIF output of a reasonably nice card like the ESI Juli@. And that is simply not the case. But it's fine if you want to believe it does, whatever makes you happy.
However, one other comment. TAS is simply ignoring this whole uproar. I got their latest issue yesterday and they did publish several letters to the editor about the article we're discussing, but the letters were clearly of the clueless class and the responses simply never acknowledge the scientific fact and logic about the criticisms of the articles posted here. Keep your head in the sand, TAS, and I will keep my subscription fees. Losing respect, really fast...
I was also surprised that the latest TAS completely ignored this. They must have received letters critical of the series. You're right, the letters they published and the response were totally clueless. To give them the benefit of the doubt, perhaps they are planning to address the issues in greater detail than the letters section allows for in a longer article in a future issue. Perhaps.
Here is a scientific approach for comparing FLAC vs WAV: http://www.computeraudiophile.com/blogs/FLAC-vs-WAV-Part-2-Final-Results
Ncdrawl,
I am not about to defend the authors as I have issues with much of the work they did.
But in terms of FLAC verses WAV, yes the theory says they will be identical but FLAC is a "protocol"...it has a written standard on how to both encode and decode a file.
This means a human needs to write software to implement the protocol and as such different execution paths will be chosen by different programmers.
Look at in another light... video files (WMV, quicktime) are also a protocol but due to human error it is possible that one video renderer will provide for a smooth and seamless viewing experience while another with the same file will be jerky and splattery.
So while you are correct that a perfect implementation of a FLAC encoder/decoder will provide a perfect WAV file, the different implementations used for encoders/decoders (due to humans) could induce issues.
I commented on this before.
If the authors felt they could perceive reproducible difference then they should have engaged a computer expert to look at and record resource utilization during use, use program profilers to see if one program was more efficient than another, see what and how many system/kernel calls were made, were there memory leak issues etc. This way some kind of correlation may have become obvious.
But to state they could perceive reproducible difference and then come up with "its unexplainable" is a terrible cop out in relation to computer software... given the decades of reseach into readily available software profilers.
Peter
And they could just have checksummed the series of files they made through WAV-FLAC-WAV conversions rather than just listening to them to see if the are similar. This shows a complete lack of understanding of even basic file management.
thats valid for the conversion process used in creating/reconverting a file... I was more referring to replaying a FLAC into a DAC via usb or spdif.
Now we are dealing with a bit stream connected to the DAC with the potential for noise/timing issues to be introduced via poor programming.
Peter
Peter,
This could indeed be, but then we are talking about conversion on-the-fly before and during playback, done by the player at hand.
Here we are talking about converting the files prior to playback, so WAV1 to FLAC to WAV2. The claim is that WAV1 and WAV2 sound different, and if you do a WAV2 to FLAC to WAV3, WAV3 is degraded even more, despite the fact that all WAV# files are bit-for-bit identical.
I must emphasize this is a completely different scenario as what you describe.
For that matter, I put the ripping-at-different-speeds in the same category... Again, the resulting WAV files from each rip are all bit-for-bit identical.
Peter
"video files (WMV, quicktime) are also a protocol but due to human error it is possible that one video renderer will provide for a smooth and seamless viewing experience while another with the same file will be jerky and splattery."
The key difference here is that WMV files are lossy compressed. With lossless compression, either the computer code works correctly and gives the correct output, or it's broken. This is all ncdrawl is saying. The real issue with this and all other "audiophoole" beliefs is the unreliability and short term nature of human hearing. Our hearing is not as absolute as many people believe. Once they understand that, audio forums will drop to 1/3 as many posts. :->)
--Ethan
This is about the Charles Zeilig article but bear with me for a minute…I’m an audiophile and so are several of my friends. We go to stereo shows and listen to many high performance systems, then discuss what we heard. Guess what, we agree on some systems and disagree on others, yet we all heard the same systems. Seldom do we agree on our fav.
Here’s where I’m going with this…Some have called out TAS as irresponsible for publishing the Zeilig article. I don’t share that opinion. Most readers can make up their own minds about this piece. Just because TAS published it along with its Wav-to-Flac-and-back controversy, one can’t assume TAS either agrees or disagrees with the findings. TAS isn’t a scientific journal; it’s a hobbyist magazine with independent reviewers and several points of view, not one party line. Heck, ever notice that TAS editors and contributors don’t always agree among themselves? Is that such a bad thing?
The good news about this Wav/Flac controversy and some other Zeilig findings is we can test them easily and inexpensively. It’s easy to compare our favorite songs and highest resolution songs in both formats on our systems. Ditto with rips from dBPA, EAC, iTunes, etc. Ditto with music playback from iTunes, JRMC, Foobar, etc. JPlay even offers a free evaluation copy. You get the idea.
If you hear sonic improvements, then congratulations, you’ve found a way to improve your system; if you don’t, then so be it, better luck with your next system upgrade.
I get it that some critics are offended that science is violated by Zeilig’s assertion that he hears degradation in Wav-to-Flac-and-back. But listen for yourself and make up your own mind. Zeilig’s ambitious effort has produced several findings to put to test, but respectfully he is just one data point in a sea of listening tests, assertions, and opinions.
There are more things in heaven and earth, Horatio, than are dreamt of in your philosophy. Hamlet, Act 1 Scene 5
I think the main point is that unless there is user error in the WAV-FLAC-WAV conversion process, there is really no possibility of hearing differences, other than ones caused by a poorly set up test environment, kind of like relying on dirty test tubes for a biology or chemistry experiment. And this is compounded by the fact that the authors either didn't know or care enough to even compare the files to see if they were the same. That part of the article and that part of their testing was just too imperfect to have any value at this point.
I get on TAS over this because TAS should have known this or they're not editorially qualified to publish computer audio articles. Whether TAS is or is not a scientific journal, had an author compared two pieces of hardware using vastly different interconnects on each, or different sources of power for each, etc., they and their community would have been all over not eliminating the right variables to have a meaningful test. But it appears they didn't know they were doing basically the same thing on the digital side.
See, this is where people need to be careful.
The main point is that unless there is user error in the WAV-FLAC-WAV conversion process, there is really no possibility to attribute any differences you hear to the WAV-FLAC-WAV conversion process itself.
Whereas one cannot reliably state someone cannot hear a difference in the playback of the two files, the article's authors cannot reliably blame whatever differences they hear to a problem with the conversion process. The same logic goes for the issue of using different ripping speeds.
> "there is really no possibility of hearing differences"
This is the key. However, people do hear differences even in an A/A comparison were nothing changed. So this proves that casual sighted listening tests are not reliable. In other words, start with the premise that Wave and FLAC sound the same, because they must, then try to understand why people think they hear a difference anyway. This is the right approach IMO.
--Ethan
Hi Ethan,
I would like to have your input (or anyone else's for that matter) on the following:
I have great trouble understanding the following quotes from Dr. Zeilig, earlier posts here:
"However, with respect to auditory perception, a thorough understanding of the implications of modern research into how the ear-brain system works suggests that double blind testing is flawed and not sensitive enough to pick up the kind of differences that most intellectually honest reviewers and audiophiles hear and hear fairly easily."
"The arguments for or against the sensitivity of double blind testing in picking up subtle (or not so subtle) auditory differences is not winable in this forum so lets agree to give this one up."
These quotes strongly suggest that Dr. Zeilig does not believe double-blind testing to be a valid way to test audible differences, but apparently this does not apply to single-blind testing as Dr. Zeilig states that all tests are performed single-blind.
Now, the only difference between double- and single-blind testing is that the person who is doing the switching and noting the results is not aware of what is tested himself to prevent influencing the listeners. As an aside, most of the testing is done with different files (as opposed to equipment); double-blind testing could have been easily arranged.
With this in mind, I fail to see why single-blind testing is not "flawed and not sensitive enough" as well, but I could be wrong here.
After thinking about this for some time, I just wonder why someone would go through trouble to set up a series of singe-blind test if one questions the reliability of the method to begin with...
Peter
I have trouble with this as well. The problem with the claim is that even if any kind of blind testing is "flawed and not sensitive," I can't see any argument for sighted testing being more accurate in some way. The main differences is eliminating a very important source of potential bias. Between single and double blind, we know with some certainty that researchers unconsciously influence subject responses based on information they are privy to; there seems no reason not to want to eliminate this bias even if you don't think it's that significant.
We also don't have much information on how the tests were carried out. Was there an ABX test showing that the subject could correctly identify one recording over another? Could the subject specifically indicate what the difference was? Or was the subject played A and B and just asked which sounds better?
And, finally, as many people have been asking here, why was there no side by side comparison of files either in the digital realm (bit-for-bit identity) or in the analog realm (looking to see if there are any perceivable differences in the waveform)? The experiments conducted by Mitchco linked above (http://www.computeraudiophile.com/blogs/FLAC-vs-WAV-vs-MP3-vs-M4A-Experi...) suggest that there is no measurable analog difference between flac and wav, for example.
> "I just wonder why someone would go through trouble to set up a series of singe-blind test if one questions the reliability of the method to begin with."
No kidding Peter. Blind testing is the gold standard for all branches of science, at least when assessing subjective issues such as taste, pain relief, and in this case audio quality. Only audiophooles believe that blind testing is flawed. I suspect Zeilig knows his logic is flawed from his statement that the discussion "is not winable" so basically let's just agree to disagree. This tells me that he knows that he has no valid argument against blind testing.
--Ethan
The problem with this position is that even in the latest issue of TAS, Zeilig and Clawsen are STILL calling for the download industry to stop using FLAC. That's a very destructive recommendation from one of the two remaining US audiophile print magazines, hobbyist or not.
I am dumbfounded at this. I was seriously hoping that TAS would at the very least address the controversial nature of Zeilig & Clawson's conclusions and perhaps publish a countervailing point of view (of which there are no shortages). But to continue blithely publishing more of this FUD against digital audio while pretending there is no legitimate controversy is bizarre to me.
I am beginning to think that some of the posters in this thread are right about a general lack of understanding regarding computer audio at TAS magazine. Audiophiles "from a bygone era" is a good description. There is a compelling argument to be made that the audiophile press has a vested interest in deprecating digital audio over analog, and indeed they have been the main resistance to digital audio since its inception as a consumer medium. But I cannot as yet conclude that this means anything more than ignorance and bias. These are powerful enough forces that there is no need to invoke a conspiracy to explain such specious conclusions.
To call on the industry to abandon FLAC -- based on these abjectly flawed articles that any of us with a computer can refute -- is beyond irresponsible; it will likely backfire. I think much of this issue is a generational gap. Younger people who grew up with computers will simply laugh at the spectacle of these guys claiming to hear differences between bit-perfect copies, especially when they say things like "double blind testing is not sensitive enough" to allow one to discern the differences that are "easily heard" by audiophiles. Or when they try to claim that your listening tests are invalid because you don't have the years of special training to know how to hear these differences...
That is the most on target description of the situation yet. Exactly my thoughts!
davedaw,
Bizarre or not, it happened. I am convinced TAS was fully aware of these articles being controversial. In fact, I would say that was the reason for them to publish, perhaps in an attempt to attract more readers. And you know what? I would not be surprised if the number of readers of TAS will indeed rise. Contrary to what many people here think, I believe there are many people who actually find confirmation of their own "experiences" in these articles and many others who believe every word written in a magazine because they assume articles are written by knowledgeable people.
I often compare the audiophile scene with the paranormal scene. Anecdotal reports suffice, and the burden of proof (effectively) lies with people who have doubts :-) But wait a minute... ABX, DBT and SBT are all invalid; they are not sensitive enough, cause stress, are fatiguing etc... And measurements can not be relied upon because they simply do not show what can be easily heard. During sighted listening, that is :-)
Anyway, as long as the audiophile scene sticks to the "believing is hearing" concept (as opposed to "hearing is believing"), we flat-earthers will always remain deaf, untrained and in possession of the wrong audio-equipment, so all the wonderful "easy to discern audible differences/improvements" will pass us by.
Peter
[quote]Has produced several findings to put to test[/quote]
No it hasn't. This stuff has been well understood since the 30s. You simply cannot get around physics and the principles of digital audio. No way, no how. Period.
Actually, it has. There are many possible hardware and software combinations, some of which will undoubtedly sound better than others. Unfortunately this series of articles does little to help listeners in their journeys, because its tests are relatively meaningless. However, the authors HAVE identified a number of potential problem areas in the area of computer music playback, and also unfortunately have MISidentified some others.
"Just because TAS published it along with its Wav-to-Flac-and-back controversy . . ."
This not a "controversial" issue, any more than claiming the Earth is flat would be controversial. It's just blatantly wrong, and it's pseudo science. There is no excuse for this, and it completely destroys the credibility of the authors and the magazine to make that kind of claim.
Yes, that is a huge fallacy, when everyone knows that your ISP is the biggest determiner of the quality of downloads. For example, once I got Comcast to provide me their best "audiophile" cable modem, and I switched to a Monster ethernet cable sitting on a soft, purple pillow (purple is key, although I hear red is also good), the sound got so smooth and silky!
***edit*** sorry I just had to vent.
I hear ya...
testing I have done has shown that for DSL, the distance to and model of DSLAM (digital subscriber line access multiplexer) has a major affect on download quality.
I have also found that by using a dialup modem using the V.92 protocol creates a blacker background when compared to DSL.
Essentially, the slower transfer rate of dialup allows my PC to use less processing power during the download and thus increases soundstage width and provides for a blacker background during playback.
I also plan to force my dialup modem into V.34 mode (33.6 kbit/s) to see if this provides for even better sound.
Peter
One thing to consider in this discussion is the error correction function computers use. For example, when you download a file from the internet, the sending and receiving computers communicate with each other to confirm the data was sent. In other words, computer sends bit A, then the receiving computer tells the sending computer that it received bit A and to send bit B and so on. This continues until the whole file is transferred. Sometimes the receiving computer may not get the data correctly and ask for the data again. This process continues for as much time as needed for the entire file to be transferred. This is the same principle for reading an optical disk, hard disk, memory stick, etc.
This works fine for data files. The problem is that music is being played based on a clock or in real time (the data is sent in a timed manner), whereas a file transfer does not need to have the data transferred at precise intervals (or bit rate if you will) – it only needs to know the data was transferred (of course there are timeouts but this is not a cadence function). ). In other words, to provide a constant stream of music, the DAC must see a constant stream of data. If there are differences in the timing of the data sent to the DAC, some timing error occurs in the DAC output which can be audible.
Think of it this way: If there is a problem with the music data, the computer's error correction function may not necessarily have enough time to sort out the data and timing issues ideally for a DAC. Timing could also be an issue if the data comes from different locations on a hard drive. Each location may have a slightly different access or read time. This could explain why a defragged drive may sound better as the file access time is more uniform. Although read access times are very fast, the read time for each bit could be slightly different if the drive has to access bits on significantly different sectors. Most hard drive manufactureres publish average seek times, not sector-by-sector seek times. And, no storage device is perfect.
Think about this - a couple CD Player manufacturers (Creek, Rega) have minimized this type of problem with reading CDs by introducing a memory buffer into which the data stream is read into and then used by the DAC.
There are still many things not measured that are generally accepted to have an impact on sound. The field of consumer computer audio is still young with many discoveries to make.
Hi bunnybeer,
I understand your line of reasoning, but it is not completely correct.
Files on a storage-device (hard-disk etc... ) are read in blocks and then transferred to the player. Your remark "the read time for each bit could be slightly different" seems to suggest you believe bits are read individually by a player, which is not the case.
Furthermore, the data-processing capacity of even moderately powered computers is enough to handle audio in a timely fashion unless there is something seriously wrong with it.
And about measurements... What kind of measurements would that be? And general acceptance within a specific group of people does not make it truth.
Peter
Edit 1:
As long as the data is with the computer's domain, all is buffered and checked at several stages. When it leaves the computer, for example interfacing with an external DAC (USB, S/PDIF or something), there can be other issues at stake.
Edit 2:
Things as sound-stage or depth can not be measured simply because this is a human spoken-out-loud interpretation of what they hear. This varies from person to person, not to mention that the vocabulary that people use to describe how things sound varies quite a bit. What to think of "the bass sounds muddy" or "the micro-dynamics are lacking". These are just words. The only way to measure is to compare the source waveform to the outputted waveform. To some extent, that will differ. Always. But as long as the difference stays below certain thresholds, it is not possible for humans to discern differences.
Both hard drives and computer music playing software also use memory buffers prior to the audio data being processed, and much larger ones than the two CD players you mention.
BUT, I think the larger point in relation to the TAS article as that there is no chance any perceived differences in sound (emphasis on the word perceived) is not attributable to any wav-flac-wav conversion or any such thing.
This refers to Charles Zeilig’s recent Computer Music Audio Quality articles in TAS.
I’m trying some of his recommendations but I’m stuck on one and looking for advice. He changed his desktop PC’s stock AC power cord for an audiophile AC power cord (PS Audio Statement, $500) and reported significantly improved sound quality. I want to try a similar cord swap with my Hewlett Packard laptop.
The problem is that typical audiophile power cords terminate with an IEC type C13 female coupler which can’t connect to typical laptop power supply units which have an incompatible IEC type C6 male coupler. The C6 mates to an IEC C5 female coupler.
Does anyone know of audiophile power cords terminated with an IEC type C5 female coupler (nicknamed cloverleaf or Mickey Mouse because of its cross section silhouette) and who sells them? Or, second best, is there some kind of converter plug on the market?
Any advice will be much appreciated, thanks.
Dear PhilCC,
Have you considered the cable between the PSU and your laptop? That is about 5 feet of "antenna" that is likely to pick up even more bad stuff than the mains cable before the PSU. Especially given the fact that is it low-voltage DC.
I have not seen them offered anywhere, but perhaps it is best to look for an audiophile-grade PSU that has good cables on both ends.
Regards,
Peter
Upgrade PSU Kit -
Thanks Peter, your suggestion about the DC wire is a logical extension of the AC power cable upgrade idea - a system is only as strong as its weakest link, that sort of thinking. The DC wire is typically captive to the PSU, so one would need to buy the whole kit. I haven’t seen them on the market either but it’s worth keeping an eye on it.
This reminds me that Chris Connaker, founder www.coumputeraudiophile.com, built his most recent all-out effort computer server (named C.A.P.S. v2.0) with an external PSU to help minimize electrical noise inside the computer case and away from the MB. But even with all that forethought he decided to use an inexpensive ($25) Casetronic PW-12V5A-L5 external PSU with stock cords as a practical matter. So, I don’t have high hopes that a suitable upgrade PSU kit exists at a cost effective price yet. Or maybe his experience indicates that just by using an external power supply, even a cheap one, electrical noise from the PSU becomes a non-issue. Not sure here.
There is precedent for aftermarket power supplies elsewhere in the audio chain. For instance, at www.audioadvisor.com, Pangea Audio offers an upgraded external power supply for the Cambridge DACMagic and Musical Fidelity V-DAC. So why not power upgrades for computer music servers? The idea being that electrical noise is not so important when crunching numbers in an Excel spreadsheet, but more so when trying to reproduce music. The nascent computer music server cottage industry just hasn’t gotten there yet I suppose, but I’m guessing they will.
Best Regards
Dear PhilCC,
You seem to have misinterpreted my post as it was meant to be read as ironical. Given my earlier posts here I believed that to be a dead giveaway, but apparently I was wrong. Sorry for not being more clear, and please don't see it as making fun of you.
I do not believe power-cords have any influence on sound quality. To me it's just another way to get money out of your pocket. Below I offer some food for thought:
There is this British audio-brand that always advised their clients to use the stock power cords because "no audible benefits would come from using other special designed power-cords". This policy changed to "We have always known that power-cords could have an impact on sound quality", but only AFTER they started selling their own EUR 450,- power-cords.
And what to think of a company that started out making cables for industrial purposes, among others for Space Shuttles Nowadays they are best-known for audio-cables of all sorts. They started a project, conducted by a third-party research institute in the UK, to develop a method to measure audible differences in cables. This started sometime in the end of 2009. So far, the only thing they can communicate is "interesting results". What bugs me the most in this is that a company that (for example) charges multiple thousands of dollars for a few meters of loudspeaker cable apparently has no way of measuring the the quality of the product. Not even for the purpose of quality-control during the manufacturing process. They (and that goes equally for other cable-companies) do not even publish the results of controlled and statistically convincing double-blind testing.
About Chris Connaker... I am also a member of CA. I respect Chris a great deal. Next to maintaining a website and keeping it a friendly environment for all, he does a lot of work for fellow audiophiles... The C.A.P.S being one of them. But Chris is also very clear in his statement his reviews and others findings are based on subjective listening, and accepts that others might have different experiences. My guess is that your are right in assuming electrical noise is a non-issue.
Anyway, it's up to you to decide where you will spend money on.
Regards,
Peter
I suspected possible irony, but I understand and respect your point of view and am not taken aback by it. Irony is a good way to get a point across.
Probably no surprise, I believe upgrade cables help but when cost/benefit is considered there is a steep point of diminishing return. But I’ll bet we have several audio opinions in common. For instance, I too believe that Chris Connaker does a fine job and really helps sort things out for the rest of us. He has computer expertise which we certainly need more of right now as the high-performance audio industry adapts to computer based audio.
At any rate, I’m going to keep pressing ahead to test an upgrade power cord on my computer music server. Zeilig’s idea fascinates me – he is the first to think of this to my knowledge, much less report an improvement. I just located an adapter ($25) to mate C13 and C5 couplers.
BTW, that was a nice bit of batch code you wrote for your WAV to FLAC 250 and 1,000 iterations testing as shown in your 2/4/2012 post. I assume you used the FLAC.exe executable from EAC? Did Zeilig respond back to you about your test?
Regards
Hi PhillCC,
There is probably also a lot on which our opinions differ :-) But that is fine!
I don't have the last of the four articles yes, so I am not sure what Dr. Zeilig wrote about exactly. But just to satisfy my curiosity, what for you is a compelling argument to look into this power-cord stuff other than Dr. Zeilig's findings (which to me are based on believing is hearing instead of hearing is believing; his FLAC & ripping findings prove that, at least to me).
I think I downloaded the FLAC tools from the FLAC project on SourceForge, but I do not recall. Too long ago.
Glad you liked the script (just 5 minutes worth of work though). But no, Dr. Zeilig has not responded. Which is not much to my surprise as he has not answered or commented on posts here since over a month. But what can he say about it anyway? That I am experiencing hearing problems? That my audio-gear is not good / revealing enough? Both?
Regards,
Peter
My two cents: while if I were a betting man, I would bet that power cords and upgraded interconnects basically do not improve anything, I am willing to keep an open mind that it may be possible under some circumstances for there to be a difference. Possibly for example in upgrading from something extremely cheap ($2 Radio Shack) to something of minimum quality spec. I just don't get all this interference and conductance crap - to me it's just as likely that a cable of higher quality metal would also transmit interference in higher quality. I look at all this stuff as basically "bling" with a dash of eliminating-neurosis-surrounding-audio-quality thrown in for good measure.
I definitely believe that higher quality source, preamp and amp can make a difference, and I believe there was also some improvement when I went from 16 gauge (please don't say it...) to 10 gauge speaker cable (although I think I would only say that it was somewhat louder, not of higher quality). I also believe that theoretically cleaner power (i.e. power conditioner) might help certain components, especially those with moving parts (e.g. turntable) although I don't believe I heard any difference when I put in a power conditioner (it's still a good surge protector).
The point is, there are certain things we can easily just call BS. But others, on a basic basis, it is important at least to keep an open mind or one is being a flat-earther. No I will never have a $400 interconnect (unless I find it at a garage sale for $10) but I did go to our local audio store and beg them to clean out their basement and sell me their used interconnects for $15 per, and I think that was just worth doing to replace radio shack, which were prone to sound dropouts when bent, stressed, loose, etc.
Only saying it is useful to keep an open mind, but by no means should you believe everything an audiophile says.
jjj666,
As for basically any type of "cord" you can go too low on wire thickness. It needs to be able to transport a signal without too much resistance, or better phrased, be up to specifications. I agree with you remarks about "better" conducting materials conducting interference better to...
I am also quite skeptical about power conditioners. The reason for this is that well designed audio-equipment should have equally well designed power-supplies. For the better audio-stuff this is mostly the case, and there simply is no need for such a device. The transformation from AC to DC takes care of that,
There is an obvious difference between a well constructed cable and a badly constructed cable. But there are quite a few cable manufacturers that are not charging astronomic amounts of money for cables (of any kind) that do just that. And one can also choose to go DIY... There are many good components available for little money, allowing one to construct cables of excellent quality.
Regards,
Peter
Hi thesurfingalien - Peter
Part 4 is subtitled “How to Make your Computer Sing, and Final Recommendations." It too is provocative. But maybe not. I’m surprised to see so few reactions or maybe I just haven’t looked in the right places yet.
I’m interested in trying this because it will be easy to swap cords at my computer music server in my stereo and decide for myself. Easy AB testing. Unlike when I was shopping for amp and speakers, I don’t have to visit far-flung stereo stores to compare. So why not give it a go? I even have access to three brands of these cords. I’m fascinated that Dr. Zeilig got so much bang from this easy swap (+35 on his sound scale) so I’m greedy to get my +35 too! (Actually I’ll be amazed if it’s anywhere near that much.)
What’s stopping me? As you know, I described a problem attaching cord to incompatible laptop power adapter coupler that Zeilig didn’t run into because he used a desktop apparently. But I’ve seen a cheap ($25) adapter on offer that may solve this problem. So if I try it and don’t hear an improvement, so be it, I’m only out $25.
What if I don’t hear a difference? Well, here’s a provocative statement on my part. I wouldn’t stop here. My computer front-end and music player software have room for improvement anyway and I’m working on that (I’m on the learning curve). Once improved, I would then redo the cable swap test. High resolving systems build on themselves you might say.
You’re probably right about how Dr. Zeilig would have responded to you. Too bad he didn’t though. He should because you presented a credible challenge to his findings. You took things beyond the word level and actually did WAV-Flac conversions and listened to the results as he did. That carries more weight than just having a bits-is-bits argument.
Regards
“In theory there is no difference between theory and practice. In practice there is.” – attributed to Yogi Berra