 
|
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 |
70.181.190.237
In the Feb 2012 issue of the Absolute Sound, Part 3 of Computer Music Audio Quality presents some disturbing findings about FLAC files. Once a WAV file is converted to FLAC and back to WAV, the secondary WAV file doesn't sound as good as the original. This has some serious implications about Hi-Res files purchased in the FLAC format.
Interesting reading.
Follow Ups:
To quote myself:
Last year I put together what's going on while ripping and/or converting to flac/wav and compared the results bitwise.
http://soundcheck-audio.blogspot.com/2011/05/cd-extraction-study.html
Conclusion:
Decrypted .flacs are 100 identical to .wav files on the PCM data
From a data value perspective there can't be any audible difference.
Since I also experience a sound difference, doesn't matter what DAC or OS or system I'm using, something else must be causing the difference.
There are certain factors which do or might have an influence on sound.
1. The flac realtime decoding process on the renderer causes
additional non linear load. That non linear additional load
translates into a pyhsical impact on the data stream.
Note: The bits received by the dac will have the same value!!
What can be different though is the shape of a bit, the timing, the
noise surrounding it.
That's what causes the audible differences on the DAC side.
2. If two .wav files sound different even though they are 100%
identical, the storage location (fragmentation or device type) of that
file might be causing a difference and some indirect effects while
reading and processing them.
Maybe even the RAM storage situation might cause a slight difference.
If the testcase is done right I wouldn't expect any differences
on 100% identical .wav files.
To eliminate most of those audible realtime decoding effects offline flac decoding is recommended.
You'd need a sw-player which decodes the entire flac to wav (PCM) prior to playback and plays those decoded PCM data from RAM (disc/cache).
All this is a workaround to cover the actual problem.
We wouldn't experience any differences, if audiointerfaces would be able to decouple from PC induced distortions, such as timing variations and noise.
The whole mess is related to the limited capability of todays audiointerfaces to cope with those physical transport induced distortions.
I found a workaround for myself. On my Squeezebox setup I decode flacs on the server.
The renderer doesn't cause any audible differences that way. I send a decoded flac or a wav stream. The renderer will get exactly the same
data and does the same processing on both base formats.
Enjoy.
Cheers
--------------------------------------------------------------------------------------------
::: Squeezebox Touch Toolbox 3.0 and more ::: by soundcheck
115 posts and not a single report of an ABX test confirming the claim (initial premise). It only takes one person and a bit of work to do such test. But I do not believe any will be published, as has always been the case in the past. Talking is much more fun.
I'm glad TAS published that article. I had a similar (but not quite the same) experience. Let me explain.....
I ripped a CD with error correction enabled using plain ole iTunes to AIFF on my Mac Mini down in the basement listening room. I wanted to play the files on my Mac laptop computer. I figured I would just transfer the ripped tunes using a USB thumb drive which I did.
The music sounded dull and lifeless when played on the Mac laptop. I tried transferring the files again but this time I used a CD-R. It sounded much better. How could that be? So that got me thinking.
I tried a variety of different brand USB thumb drives including Kingston, SanDisk, HP, LaCie, Patriot, Lexar, Transcend, etc.
I couldn't believe my ears! Some transfers were warm and smooth, others were detailed and very transparent, some were actually distorted and unlistenable. The LaCie thumb drive transfer sounded best to me but this was to be expected as LaCie caters to the Mac crowd with lots of Mac compatible products.
Have I discovered a new niche within this audiophile hobby we all enjoy? We need a "Thumb Drive Rolling" forum for folks who enjoy trying out different thumb drives much like the tube rolling crowd.
I'm going to dig up my NOS collection of early thumb drives. I bet they sound better than anything currently in production.
I guess this means you are all thumbs...
nt
Cut-Throat
There's been a phenomena that's been bothering me in my system for several years now. Almost all of my music sounds good, but some of the hotter masters seem to have unnecessary and unnerving distortion. I hear it on kick hits, on deep snare hits, on the bass guitar, etc. We're talking really hot masters, like John Mayer and Coldplay. The single worst-offending song is Clarity on John Mayer's 2003 album Heavier Things (starting at 1:21).
I've tried two different pairs of good headphones, two different headphone amps, two different integrated amps, three sets of speakers. I've tried three different media players, I've tried using Windows sound vs ASIO4ALL vs WASAPI. I've tried three different DACs. But I can't seem to get the sound to go away. I thought it was just in the music, which was irritating, until I listened to the actual disc on a good system. The distortion was gone. And believe me, I was looking for it.
What I find interesting is that I told my brother (a non-audiophile) about the phenomenon and played him Clarity, and his nose wrinkled in disgust. But then I played it for mastering engineer Bob Katz on my system, and he couldn't hear it. But then again, he was driven batty by some sort of "distortion" or "digititus" on a certain piece of very high-end outboard DSP he was using, and I found this distortion to be completely inaudible. I guess different people zero in on different things when they listen.
The last paragraph aside, I really can't think of many pieces of the signal chain that remained constant when I was switching out gear to isolate what was causing the distortion, other than the fact that a computer ripped it, it was being played from a computer, and that it was converted to FLAC. I have believed the conversion to FLAC to be the culprit, although I'd like to do some more testing to be sure.
I suppose the next step would be to test on my new system that has new speakers and a new DAC. Assuming I hear it, I'll play the CD in a cheap CD player connected to the system. Assuming I don't hear it then, I'll play the CD on my computer. Assuming I don't hear it, I'll play the original WAV file (I keep my entire library in uncompressed WAV as well as FLAC level-8). Assuming I don't hear it, I'll have my answer.
Has anyone else experienced this type of distortion? Does anyone else have the song Clarity by John Mayer to compare? Is this exactly what Absolute Sound is talking about or totally separate? (I'm not a subscriber, unfortunately.)
Chances are the really hot masters have horrible distortion on them. Many mastering engineers are so used to this distortion that they no longer pay much attention to it (in the mild form). Some of them make their living butchering music so that it will be "loud", which is really stupid since all this does is save a slight clockwise twist of the wrist.
Bob Katz has fought against the loudness wars and I would be surprised if he didn't hear this distortion. But then I haven't head much of his new work. Some of his older recordings, such as two of his Ana Caram recordings on Chesky, have superb sound. But perhaps he, too, has gotten used to doing what is necessary to earn a living in today's sick music industry.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Bob is not in the habit of clamping down the limiter beyond all reason. I've seen him counsel clients towards more dynamic masters, which I approve of. I've also seen him reject clients because they wanted a louder master still, even after he put significant hours of labor towards their music. In this latter case, after explaining the costs and having the clients acknowledge and ask for a louder master, I'd just do it. It's not as if music designed for the clubs is meant to be a breath of fresh air, anyways. And the client is always right...
I repeat, Bob did not detect this distortion problem on my system, even though I pointed it out to him and played the worst offending track for this problem that I know of.
But when I used the actual disc on Bob's system, I did not hear the problem. I really should try to hear it in more people's systems, but it is not one of my test tracks. (Perhaps it should be simply for this reason?) Also, I don't have a great deal of exposure to high quality systems.
There is another possibility, sad to say one that I've experienced. I have no idea if that is applicable in your case. (I hope not.) The problem is an interaction between a recording, a system and a set of ears.
In my case, I noted that certain recordings (e.g. percussive piano attacks played at piano in the room volume) would cause distortion, but only out of the left speaker. My first impression was that this was due to a defect in the tweeter. I swapped the speakers left to right. The problem remained in the same place in the room, but now was appearing out of a different box. So I swapped channels left to right. The problem remained on the left, so it wasn't on one channel of the recording or one amp, etc. My last shot at identifying this was to spin around in my chair so that I was facing in the opposite direction and have the sound come from behind me. Now the "distortion" was still coming in via my left ear, even though this ear was closer to the other speaker. In other words, this was a problem in my left ear, not in my system. It only happened when playing some recordings at high volume levels. The problem also went away if I simply moved further away from the speakers. Apparently the speakers were putting out some sound that my left ear was hearing as distortion. By backing out of the room and listening from a distance, this system will play cleanly at least 10 dB louder than if a full size grand piano were in the room, so the problem is lack of headroom in my ears, not in the system.
I had a similar experience at a fund raising dinner for the local operate company when a Soprano was wandering around the tables while singing and got within about 10 feet of me. I could hear my ears distort. I don't recall this problem previously, perhaps this is the effect of age.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
a Soprano . . . got within about 10 feet of me. I could hear my ears distort. I don't recall this problem previously, perhaps this is the effect of age.
Hmmmm. Your age or the soprano's?
D
"Your age or the soprano's?"
Go figure. She was about one third my age. :-)
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Huh, that's something to think about. Thanks.
Even if this is true, who cares? Most people I know cannot, for the life of them, hear any difference between lossless WAV and crunched mp3.
I've compared the original CD .vs .wav .vs FLAC and there is no audible difference between the 3.
AIFF seems to be a different story. I have never heard an AIFF file that sounded like anything other than shyt.
http://www.AudioAsylum.com/forums/pcaudio/messages/10/101742.html
While I seriously doubt that encoding to FLAC could cause a permanent loss of quality, I think it's possible that it does sound inferior to WAV during playback. I've noticed that at least some of my FLAC files are Variable Bit Rate. Is it possible that decoding the VBR file introduces more jitter than when decoding the Constant Bit Rate WAV file?
"I've noticed that at least some of my FLAC files are Variable Bit Rate. Is it possible that decoding the VBR file introduces more jitter than when decoding the Constant Bit Rate WAV file?"
I agree with your comment about variable rate CODECs potentially introducing jitter. They require an extra level of buffering in the playback program and there will be more frequencies involved in the buffer reloading processes, so if the underlying hardware (computer, sound card and DAC) are not excellent there will be jitter introduced by the more complex data flow.
All FLAC files are variable bit rate. Indeed all lossless CODECs are variable bit rate. This can be proven mathematically:
"Lossless data compression algorithms cannot guarantee compression for all input data sets. In other words, for any (lossless) data compression algorithm, there will be an input data set that does not get smaller when processed by the algorithm. This is easily proven with elementary mathematics using a counting argument, as follows:..."
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
I thought it might be possible to put uncompressed audio in a FLAC container. While this wouldn't allow compression of the file size, it would allow for compatible tags. Thanks for the reply and the link.
I believe that it is possible to use uncompressed FLAC. This is just a matter of selecting FLAC level 0 compression. I haven't checked it out. I believe it will increase the file size slightly for the needed header. Whether it improves sound quality is another question.
There is no reason not to put tags in WAV files. There are many software packages that seem to handle tags correctly. Unfortunately, not all do.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Interesting discussion http://lists.xiph.org/pipermail/flac-dev/2011-May/002950.html
Thank you for the link. I found Mr. Willoughby's comment, that if FLAC, WAV and AIFF don't sound the same then something is wrong with your system, most interesting.
Very nice link. Obviously inherent in everyone's knowledge (there) is that WAV> FLAC> WAV conversion is transparent. The question is whether real-time playback of FLAC might differ from WAV, but I couldn't find any actual testing of that hypothesis, merely some opinions that it might, opinions that do have some supporting data in other places (some of Gordon Rankin's posts here at the asylum, for example).
One of the first points is that CD ripped to FLAC sounds worse, although there is no clear explanation of how playback is effected or the method and scope of assessment.
Then it goes onto FLAC and player plus FLAC and hires.
One needs to read the thing carefully to determine what one thinks it is saying. Your caption is somewhat too deterministic.
In this repect, it is a poor article and carries the same low standard of 'hifi audio' articles.
I hoped that others would read the article. This was one finding that the article presented. I just wanted to initiate discussion.
Edits: 01/07/12
reacted without even reading it. No way to make progress.
nt
Cut-Throat
serious posters to get hold of it, read it, and then comment;certainly no shooting from the hip.
nt
Cut-Throat
.
But IMO, that alone does not preserve playback fidelity.......When a signal is first digitized, the jitter signature of the original digitized signal is *not* perfect. If the original jitter signature is not altered, I think the playback will yield the best fidelity. But if the jitter is altered, or even lowered, the deviation from the original jitter signature degrades the sound.
If lossless compression were to be applied, the jitter signature would be totally changed. The jitter might not necessarily be worse (as measured), but just changed altogether. This could explain why lossless compression algorithms such as FLAC does not satisfy some listeners.
Edits: 01/06/12
Any jitter on the original analog to digital conversion was encoded into the sequence of digits in the original file. These digits would have been different if the jitter on the A/D clock had been different.
This distortion on the original recording will be forever preserved by any sequence of file copies, however performed. If one of these copies is played back with equipment that has its own jitter problems then the playback jitter will be superimposed on the recording jitter, and there is little chance that this will do anything but make matters worse.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
> Once a WAV file is converted to FLAC and back to WAV, the secondary WAV
> file doesn't sound as good as the original. This has some serious implications
> about Hi-Res files purchased in the FLAC format.
It has more implications for the article. I have done this and both files are
bit-identical. As they are bit-identical, the bits presented to the DAC are also
the same. So if the authors heard differences, the only reason could be due to
differences is disc activity due to the 2 files being fragmented differently.
Which is a factor that could go either way.
John Atkinson
Editor, Stereophile
I agree. This is the one thing that I do not agree with in the article. FLAC files that I have converted back to .wav sound fine.
The FLAC files dont sound fine however. Has nothing to do with doing static data compares IMO. This is a real-time effect.
Steve N.
As Mr. Atkinson adeptly observes, the files are the same, 1 is always 1 and 0 to my understanding, maintains that position.
The implied difference is that the Absolute Sound sold more copies this issue, than last.
Data compares done statically have no bearing on how FLAC sounds dynamically, real-time. FLAC is broken IMO.
"FLAC is broken IMO"
Steve,
Is it really that extreme? Or is it possibly small perceptible differences that some might be able to hear if they are being super critical?
regards
Bob
Not to me, in my system. It sorta seems like WAV might be a bit more open and immediate sounding but the difference is so small that I'm not certain that it isn't just my imagination. I'm going to compare a few more times to convince myself one way or the other. That, I think, it what each of needs to do rather than simply accept what we read.
You mean your wife can't hear the difference when she is in the hallway with the door closed? :-)
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Your smiley at the end did not escape me but your meaning did. Are you suggesting the WAV is so superior that you'd have to be listening under extremely poor conditions to not detect the difference?
Speaking of my wife, I am *very* thankful for her. The last time she sat in the sweet spot (5 years ago!) she proclaimned my new speakers to be better than the ones they replaced!
I think you understood me quite correctly. It was the other posters who heard extreme differences that might not have gotten it. :-)
You are a fortunate man to have such a wife. My late wife more than tolerated my audiophile tendencies, but she was a musician and didn't care so much as to sound quality. One day after being subjected to several amplifiers, including a 100 wpc Bedini model that Clark Johnsen had lent me, she allowed that she didn't really care about any differences between amplifiers, she had tried to listen to the sound and got distracted by what she really cared about, which was what the musicians had been thinking while playing.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Until I heard my FLAC demo myself at RMAF, I was skeptical about the magnitude of the effect. I had a customer come in the room and I told him what I was doing and he sat down and had a listen. We were both shocked at the magnitude of the difference. Like I said, FLAC was like listening through a tunnel. The customer stood up and said in disgust: "I dont want to hear this, all of my files are in FLAC". It was that obvious.
Of course, our system was very resolving and live. Our room got a best sound of show from TAS, same as last year.
Steve N.
My guess would be both. Making a bold pronouncement like "FLAC is broken" based on these types of test results is just plain stupid, however, again making me question why ANYONE would believe a thing you say.
Since there are many "best sound of show" awards, can you tell us which room was yours?
Iris room:
http://www.avguide.com/blog/rmaf-2011-report-digital?page=2
Like I said, FLAC was like listening through a tunnel. The customer stood up and said in disgust: "I don't want to hear this, all of my files are in FLAC". It was that obvious.
This is not a version of this oft-repeated tale we've heard before; usually it's been a report of a "survey" of customers visiting the stand at RMAF.
We were both shocked at the magnitude of the difference.
It was new to you at that point? You did this research at a trade show ? Is that a research technique you'd recommend to others?
Just curious.
I'm too damn busy designing, ordering parts, filling orders, packing, shipping and answering questions on email and by phone to be comparing everything to everything. I have to sleep sometime. I only have the opportunity to shave about every 3-4 days.
I already knew that FLAC didnt sound great, but this was the first time that it was really obvious. My partners in the room told me that ALAC, AIFF and FLAC all suffer from these effects to varying degrees, so they actually converted the file for me and asked me to try it.
Steve N.
If you had said that the FLAC player, the computer and the DAC did not sound great then I would have agreed with you. But to say that FLAC didn't sound great, I will not accept that. There is no basis in attributing any sonic problems to the format. There are other players that completely avoid these problems when playing FLAC as they do not reference any FLAC data or execute any FLAC instructions while the music is playing.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Apple guys please revolt.
The other day I compared an AIFF file with the same Wave file taken off the Reference Recordings HRx disc. No FLAC involved. I couldn't hear any difference between these two files playing with Pure Music.
I came to the conclusion that my equipment stinks.
I agree with Tony.
Possibly the problem is not in the FLAC itself, but in the playback method.
I mean, if FLAC is decoded on-the-fly, like most software players do, it might enduce electrical problems in the playback system.
A test could be listening in sequence to a WAV original file and FLAC made from it. Then, repeat the test with original WAV and WAV made out from FLAC.
Once again a voice of reason ...
You guys can just go on using FLAC. Not me....
I should think you would want to make it your business to build DACs that will sound good regardless of format. That way you can capture the value (revenue) the customer is currently spending (wasting) on audiophile grade software and computer cables.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Look Tony, Im trying really hard to fix this situation, but it requires cooperation from either Apple or Microsoft. I'm working with Microsoft right now. I have a proposal in to them. These things take a lot of time, if they ever even materialize. It has to be beneficial to the bottom line, or they just punt on it.
Steve N.
That would be nice in an ideal world. Its like saying "build a DAC that will sound amazing with the Windows audio stack".
Impossible I'm afraid. I dont have any control over the digital stream coming out of the computer. Once it's compromised, there is no going back. all I can do is make recommendations based on my and my customers experiences.
Steve N.
There is absolutely no reason for a DAC to know anything about the Windows Audio stack. If it has this knowledge, the battle is already lost.
If there is a single specification that is relevant, it is the four page I2S specification. If the Windows audio stack sends the wrong bits out of the box it's not your fault. But this is something that is easily verified by looping the bits back and recording them or sending them on to a second computer.
Messing around with USB is just making things too complicated. You will spend too much time worrying about M$'s cybercrud and not worrying about what matters, namely lean, mean and clean digital and analog circuitry. If customers must have USB, then let them use a converter to go to the DAC.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Exactly! This is what manufacturers should be thinking...If indeed there is degradation, FIX IT and we'll buy it. Use ad space to trump the company's achievement.
"Introducing the NEW SuperDuper DAC Mark IV! Engineered to accept FLAC/APE/WV/AIFF/WAV with NO deterioration in sound quality between lossless & VBR file formats!!!"
I honestly find it hard to believe a lossless decoding process that takes barely 5% processing power on most computers these days can make so much audible difference that it can't be fixed!!!
Edits: 01/09/12
On my computer decoding a 44/16 FLAC file and converting it to WAV, combined, took less than 0.5% of the time (greater than 200x real time including formatting and saving a WAV file). This is less than 1/10th the processing power required to upsample, so if the FLAC conversion were doing damage it's highly likely the upsampling would be also, and therefore any steps taken to improve processor performance or prevent processing from polluting a crummy DAC would apply 10x to the upsampling process. (Numbers are for an Intel Core i5 650@3.2 Ghz processor).
If one believes that FLAC degrades sound, then one can still save all one's library in FLAC format, as it will take only about 20 seconds to convert an entire album from FLAC back to WAV.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Is this trouble really worth it, just to cut the filespace down by 50%? Disk is cheap!!
There's no trouble. Just use the right playback software and there will be no remnant of FLAC affecting the sound.
Disk space may be relatively inexpensive, but the more space that is used the more spindles one has to buy and install and the more time one spends making backup copies. If one runs out of space on disk one has to buy additional boxes, etc... It's a matter of "waste not, want not".
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
The initial articles could be argued with but at least they put good stakes on the ground climbing in the right direction. Now, this 3rd article is a major setback that -- as far as I am concerned -- kills their credibility.
With a title like "Is FLAC a Fraud?", I should have seen it coming.
These guys really bit more than they can chew, grandly. And not once did they feature -- or even mention -- a bit-level comparison...OF ANYTHING!
Quote from the end:
"To answer our own question, we do not believe FLAC compression is
fraudulent in the sense that it is not truly a lossless format, but
we do believe that for reasons yet to be fully explained, use of
FLAC compression can be associated with very significant sonic
compromises."
Moronic & INSULTING! Gents, you either FULLY explain your point and reasons, or stay away from the subject. There are reasons why FLAC can be prevented from delivering its full benefit but it simply is not an issue of it losing the information contained in the file.
I guess you missed the claim in Part II about 2 identical files (but ripped at different speeds) sounding different?
nt
Cut-Throat
This is what I have been saying for a year now. Maybe the non-believers will pay attention now.
Even AIFF does not sound as good as .wav according to some of my customers.
I only do .wav files. I dont trust this flakey software.
Steve N.
"Even AIFF does not sound as good as .wav according to some of my customers."
Have you been able to hear this yourself? Are different people sensitive to different things?
I have not heard this difference myself, but I have not tried either. Maybe I'll give it a try soon when I have some time. I believe my customers. I know their systems.
I wouldn't get too attached to the container labels, .wav or .aiff. Both of these formats have a lot of parameters that describe how the files can be stored, things like the byte order (big endian or little endian). Any observed differences could be a function of the way software implements these parameters and not the outermost container.
There is no magic involved here, if there are two versions of a recording, one in .aiff and one in .wav one can use a Hex editor to ferret out exactly what is in these two files. IMO it is irresponsible to reach any conclusion as to which format is "better" until one has examined what is actually in two files that appear to sound different.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
But we're not talking about playing FLAC vs. playing WAV or AIFF, we're talking about the codec activity, and I don't think there's any question that FLAC is transparent in that way (despite these authors' claims). At least they are consistent, though; they consistently claim that two files which are identical in every way (md5, null test, whatever) can sound different based on some other "magical" quality.
FLAC colors the sound. Makes it like listening through a tunnel. Its more like compression. Its the real-time behavior that is broken, not the staic behavior. No amount of A/B file compares will show this.
The results of this study are consistent with my findings. Only .wav preserves all of the music.
In fact, the controversial part of this article is the claim that the "static" behavior of FLAC is broken. Whether WAV, AIFF or FLAC sounds better in real-time playback is another issue entirely.
as evidenced below
How can listening to tracks test the "static" behavior? This requires A/B data compares. Has nothing to do with listening.
Hmmm, I haven't read Part 3 but do have Part 2 which contains a number of questionable assertions as well. (The whole subjective sound quality scoring these guys do seem totally ridiculous anyways.)I seriously wonder whether the agenda here is to increase the amount of obfuscation and FUD around the "complexities" of computer audio.
However, digital brings with it 2 inconvenient truths for the industry:
1. A BIT IS A BIT. Anyone with understanding of how WAV, FLAC, APE, WV works knows that bit-perfect is bit-perfect. This is irrespective of how it's played back or what equipment it's stored in. There is no "jitter" in the file or other variable to lead one to question the integrity of the stored audio. It's as good as that CD or hi-res download is ever going to be.2. PIRACY. 100% copies are easily done. Exact replicas of the original. Something obviously IMPOSSIBLE in analogue.
I think most folks on the street know this (and probably don't care) but the industry will fight in any way it knows how - especially among the vocal and possibly more affluent audiophile crowd. Ultimately there's money on the line. With the resurgence in articles and interest in LP, why not pump up analogue again?
I believe that the reason TAS runs articles like these is NOT to promote computer audio. They want to sell something. To put doubt into the minds of the average consumer (perhaps the older analogue crowd?) about the simple truths around digital... Sure, you can pit one DAC against another, or consider jitter in mechanical playback devices like CD players, even jitter in digital interfaces like TosLink/coaxial/AES - how much of this is significant is left to the audiophile's ears and imagination. However, once you start questioning the *digital storage format* itself (FLAC) which is designed to be bit-perfect and has not been shown otherwise in the last DECADE (as if FLAC is anything new!), this IMO is either denial, miseducation, delusion or outright fraud.
Having said this, I still buy the occasional TAS off the shelf (like the Product Of The Year issue last month) to see what's new and to be entertained, but in terms of credibility in technical matters? I've left that camp years ago.
Edits: 01/06/12
Interesting.
What did you think about the "perfect sound forever" CD mantra in the 80s?
Not asking to be antagonistic, and despite my moniker I AM currently immersing myself in the world of computer audio. However, novice that I am, I have a hard time accepting that "lossless compression" is anything but an oxymoron.
Again, no disrespect intended, trying to learn.
"However, novice that I am, I have a hard time accepting that "lossless compression" is anything but an oxymoron."
Not an oxymoron. It's been done for decades with computer data, as in .zip files. Lossless compression works because typical files contain redundant data. Ordinary zip files will compress audio by a certain amount, but special compression schemes tailored for audio will do a better job, such as FLAC, ALAC and WMALossless.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
So you don't think that ANY manipulation of the original file, no matter how big or small, has the potential to corrupt the ultimate sound quality?
Is copying the file "manipulation"? Because that's essentially what we are talking about here.
There isn't any manipulation with lossless compression. After a file has been compressed and then expanded the result will be a file that is an exact copy of the original file. (Neglecting any issues with headers that may have been changed by other software that is not part of the compression process.) What this comes down to is that the net result of a lossless compression followed by an expansion will be no different than a straight file copy.
Whether an original file and a bit for bit copy of the file sound the same is a separate question, having to do with the computer system hardware and the operating system and player software. It has nothing to do with the compression/expansion processes, since these processes will not be in use when the two copies are played.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
before you make accusations.
Whilst I find the methodology opague and the scores questionable, there is no reason to manufacture an agenda behind the articles.
Goodness knows, there is plenty of agenda behind some of the posts here, especially when most have not rerad the articles.
Look fmak,
If mercman's summary is true (that they believe the FLAC decoded WAV is inferior), and they're touting this to the public, then it's total nonsense. If you can provide more details to refute the above then please enlighten us.
I've already read part 2 which I found forgettable and unnecessarily filled with pseudo-objective tables. Hey, I might actually buy the magazine this month to have in my possession proof of their insanity (turn into the National Enquirer to sell a few more mags).
I've read the articles and sent a letter to the editor (which they might print next issue - ha hah). Unfortunately, I think it's entirely possible (as I've posted in other threads on this topic) that they are trying to undermine the entire field of computer audio, and of course there are many reasons that might be so. I'm not saying that is the case here, but it's not far-fetched to think it might be.
Why would they want to undermine computer audio? The more commonly used formats we have, the more products we have...and more products = more advertising $$.
NT
I really hope they print your letter Rob!
In fact, I really hope they get inundated by letters!
If you recover the original bits correctly, I don't see how that is possible?
One thing I have noticed is that if your computer is chugging away at other programs, or has turned down its clock speed due to high temperature right afterwards, you can get more jitter.
It is possible that FLAC has less jitter than WAV due to the smaller bandwidth to transmit it, too.
I'll have to read the article - DOES sound interesting!
"Knowledge is knowing that a tomato is a fruit. Wisdom is knowing not to put it in a fruit salad"
I've always assumed that FLAC was a mathematically lossless encoding. So I checked, to be on the safe side of science. Well, it is. A FLAC rip (for instance of a 16/44 CD) will enable the user to reconstruct the CD identically (i.e in 16/44 WAV).
Quote from Wikipedia: "FLAC uses CRC checksums for identifying corrupted frames when used in a streaming protocol, and also has a complete MD5 hash of the raw PCM audio stored in its STREAMINFO metadata header. FLAC allows for a Rice parameter between 0 and 16. […] If a CD is read and ripped perfectly to FLAC files, the CUE file allows later burning of an audio CD that is identical in audio data to the original CD, including track order, pregaps, and CD-Text."
Of course, the same would apply to other digital sources (DVD-A, SACD, Studio, etc.)
But it's true we're not talking solely about science, because if both music and computers are understandable in mathematical terms, mics, amps and speakers are not. Not entirely. It's about technology and how it reacts to different data encodings; beyond that it's about human ears. This reminds me of the whole SACD discussion. SACD use DSD files at 1bit/2,8MHz resolution. Is that, to one's ears, noticeably better than, say 20bits/96KHz? (both are approx. equivalent mathematically, people usually rip SACDs to 24/88.2) When you actually compare the two, most people don't hear a difference, and a few will argue that A or B is better based on technical audiophile considerations.
That is, differences based on specific hardware, based on different dedicated technologies. Typically what memory play, by removing many mechanical artifacts, may help improve a lot.
The matter of the fact is, processing data through a linear tech chain produces all kinds of "perturbations" (jitter, loops, artifacts, etc.; this is much easier to observe with video data for instance) that can be better-handled through improved designs; but such improvements are usually dedicated to a particular encoding/resolution in the digital part of the chain (it's quite like amps/speakers: they don't all do well in all music genres, they have "preferences" or rather we, humans, do). There is a notion of "tailored components" in audiophile research, that's what the SACD 1-bit standard was all about.
So, while I cannot for one second believe that FLAC isn't "lossless" in the purest, mathematical meaning of the term, it is actually possible that FLAC files do not sound exactly the same as WAV, ON A PARTICULAR SOUND SYSTEM (that will undoubtedly vary from one hardware chain to the other). That's because the real world is analog, unlike digital files. And we, such as hearable music, exist in the analog reality.
However, rigorously, making comparisons between lossless encodings from the same master (say, between DSD, WAV 24/96 and FLAC 24/96) would require, at the very least, an Audiophilleo (outstanding USB-to-S/PDIF "rubbish filter" ^^) to eliminate all the jitter that is produced by data treatment within the computer (but not symmetrically between a WAV and a FLAC for instance, because, well, they're different to process…), and a serious analog section (amp + speakers/headphones). I'm not sure one can hear the difference without spending 5-figures and listening very, very attentively. And even then, it might just be a specific technology talking, doing its mojo on one format more than the other.
My advice (will follow it myself): keep your FLACs, but to get the best out of 'em:
• Pay attention to USB jitter (again, Audiophilleo1 is a computer audiophile's best friend as we speak), that will do much more justice to your D/A conversion section than nitpicking about the encoding.
• Make it short from source to speakers (ideally, FLAC > Player > USB+Audiophilleo or S/PDIF > DAC > Pre-amp (if not integrated or not satisfactory inside the DAC, but don't let the preamp taint the sound, any "Direct" mode is your cue) > Power amp > speakers).
• Pay much attention to all analog cables (bad quality here will kill your music more than any compression, because it's random, it's not even a logical loss that occurs, and there's no such thing as a perfect analog component, so your money is always well-spent here).
• Oh and get a good player such as Pure Music or equivalent.
While others will be busy re-buying or re-ripping their lib in Wav for no obvious improvement , you'll be thoroughly enjoying good lossless HD music.
How do you tell a good quality cable from a bad quality cable?
Cut-Throat
Next step should be scientific article stating that the same value of bit might differ depending on the bit in the file. Soon we will start to discuss audio involving quantum mechanics
The big topic on here was that two files that subtract to ZERO in a wave file editor (bit for bit exact) could sound difference due to some file header / offset amount. Update: File HEADERS - different header information is supposed to affect SQ. Read here:http://www.audioasylum.com/forums/pcaudio/messages/10/101564.html
Sometimes I think there are a few around here who want to be the next guy to discover some never-before considered sonic difference and get attaboys and accolades from the guys who'll drink any color of kool-aid.
Now the DIY USB cables are pouring in, each with a blacker background, lower noise floor, another veil lifted....
Ah, life in the asylum. Never a dull moment. A few deja vu's. But never dull!
Cheers,
Presto
Edits: 01/06/12
You find an incompetent mumbling of the Bozos interesting reading? You have low expectations.
It was a bit of sarcasm on my part.
It was a bit of sarcasm on my part.
A late friend of mine and that rarest of beasts, a decent politician, used to tell me never to use irony in a public forum; "they just don't get it".
It seems you haven't learnt any more than I did . . .
D
"never to use irony in a public forum; "
is using sarcasm in written form such as the internet and e-mail. I have "freind"/professional colleague who barely speaks to me anymore after he misinterpreted what I thought was plainly over-the-top sarcasm. I certainly learned to be careful of what I write electronically - unless it's AA where all bets are off ;^)
unless it's AA where all bets are off ;^)
I think you have that the wrong way round. You think an inmate is being sarcastic or ironic then find that he (and it's always, always a he) believes every word he writes with a passion that's almost frightening.
(I could of course spoil what little by way of a joke there is in the above by inserting a smilie for the benefit of those too dumb to see it's a joke . . . )
Because it happens too often.
One can't cry wolf every other day and then cry foul when someone else starts doing it.
Bed, meet crier. Crier, meet bed. The rest is obvious.
Bed, meet crier. Crier, meet bed. The rest is obvious.
The rest might well be obvious but your aphorism isn't. I haven't a clue what you're on about. Sorry.
Why does it seem like everything they write at Absolute Sound is highly questionable, at least when it comes to Computer-based Audio? It seems like every quote or link I've followed here over the years from them is written by someone who is generally ignorant about basic concepts related to digital and/or computer technology.
I read through this discussion on the TAS/AVGuide forum and agree that an incredible amount of quasi-technical, speculative BS is evident.Then, the thought hit me that none of these people ever become obsessed over how many times a digital file is transferred or copied in the production of CDs, or the uniformity of media over the life of the stamping process, etc.
Even more so, I don't recall concerns about the incredible inconsistency in the quality of analog storage media. The steady deterioration of pressing quality from the start of a stamper run, variable vinyl purity, what generation of the master tape was used for which pressing (often an issue with LPs "back in the day") and on and on are ignored.
A purchased LP that is many multiple mechanical copies from the original acetate is treated with as some inviolate, perfect object. Meanwhile, a FLAC/WAV conversion that passes a MDSUM check as a perfect copy over the range of a hundred million bits or more is subjected to endless obsessive-compulsive navel-gazing.
This hobby is always an interesting study in human nature.
Edits: 01/05/12
I long ago tired of the COMPONENT OF THE MONTH CLUB that all of the "audiophile" publications have always been but in their defence TAS has had numerous articles about LP quality.
The first place I read about looking for certain stamper runs, etc. This was easily twenty years ago.
At their best these publications report what they hear. Nothing wrong with that. We can all speculate what is the cause of differences in sound but when we pontificate with assumed authority we can quickly make fools of ourselves.
Digital can be every bit as mysterious as analogue so we should not assume that technoology has all of the answers. And please, do not confuse technology with science ...
Even if you get a LP from a desired stamper run (which is difficult to do without opening the record to examine the codes between the label and run-out groove), you still have the issue that quality difference between the start of that run and the end of it will be significant. You also have the issue of consistency of the quality of the vinyl blob used for any pressing, whether at the start, in the middle or end of the run.
I like LPs and have heard many fine ones since becoming interested in this hobby almost 50 years ago. However, the more common experience I had was one of mediocrity with periodic episodes of horridness. Since I finished converting to digital the last of my roughly 2000 records a year or so ago, the only vinyl I deal with these days is the odd LP I stumble across that was never issued on CD (Domingo Bethancourt, anyone?).
I miss nothing about the inconsistent quality of the vinyl world but would still contend that most modern-day LP enthusiasts ignore or forgive these defects while obsessing over the supposed microscopic difference between an optical CD file and its rip.
But, that is mainly just human nature. We willingly forgive the deficiencies of things we favor while maintaining a hypercritical stance toward things we don't like.
When we had the audio show in Atlanta I heard some horrendous Lp pressings. So bad it sounded like mistracking and not slight mistracking or a simply completely worn our pressing.
I made mention to the proprietors and my concerns were dismissed.
I listen to both digital and analogue and have found it harder and harder to tolerate all of the extraneous noise on the usual LP. Sure, about 25% of my disks are virtually silent but most have noises and distortions to them that easily exceed any of digitals errors.
Of course, this has become the case as we have learned so much more about digital and the SQ has made great leaps.
I suspect we agree that analogue for many has become a fetish.
Both approaches have their strengths and benefits.
If they took the time to take an MD5 sum on the original file and a compressed-> uncompressed version they would have found the results to be identical.
I encourage you to try it.
"If they took the time to take an MD5 sum on the original file and a compressed-> uncompressed version they would have found the results to be identical."Unfortunately there are some inconvenient facts that may confuse things. I will describe a little experiment I did tonight. (I've done this before, but I wanted to do it again to make sure I got all the details exactly right before posting.)
1. I started with a 96/24 WAV file that I had previously made using Soundforge 10c. This file was pure, i.e. it was made from a digitization of an analog waveform played on a tape recorder. The recording had never been "polluted" by being passed through any evil "lossless" compression. Call this wav file FILE 1.
2. I took FILE 1 and used dBpoweramp and compressed this using FLAC level 8. Call this FLAC file FILE 2.
3. I took FILE 2 and used dBpoweramp and converted this file back to WAV. Call this wav file FILE 3.
4. I took FILE 3 and used dBpoweramp and compressed this using FLAC level 8. Call this FLAC file FILE 4.
5. I took FILE 4 and used dBpoweramp and converted this file back to WAV, Call this WAV file FILE 5.Here is what I observed. Files 1, 3, and 5 were all WAV files of identical length. Files 3 and 5 had identical MD5 checksums. File 1 had a different MD5 checksum. However, by differencing the audio samples in Files 1 and 3 I verified that the audio samples in files 1 and 3 were identical, i.e. the files differed only because of the headers. It looks like dBpoweramp likes to write headers one way and Soundforge a different way. What this shows is that different programs can produce different WAV files that contain identical audio samples. This is not about FLAC compression, it's about any audio software and also about the WAV format that allows the same audio samples to be expressed in different ways.
If a difference in WAV file headers affects sound then it is because of poorly designed player software. For example, memory playback software would load an entire file into RAM memory. Files 1 and 3 would load identical contents into RAM memory. At that point the playback software will then be running on identical data in identical memory locations and the result is going to be the same. If non-memory playback software is used then while the music is playing the software will be occasionally operating on different headers and the audio data may be in different memory locations, and this may affect cache hit ratios and other aspects of program execution that might make the sound different (possibly better or worse according to some individual's taste).
If there is BS here, it's on the part of people who don't know enough about computers to understand how they actually work. I didn't read the article, so I have no comment on whether this statement applies to the author. Of course, if the computer had been connected to a DAC that was properly designed then the poorly designed player software wouldn't matter either, since the DAC would produce the same output regardless of the quirks of player software. As the quirks of player software and DAC sensitivity will continue to exist with or without FLAC compression this brouhaha has nothing to do with FLAC and all about crummy player software or DAC hardware.
As someone who makes recordings, I would find it completely unacceptable if two WAV files containing the same samples sounded differently. I am deliberately changing the samples in the file to make a recording sound the way I want, and it would be completely unacceptable if the recording were changing for other reasons. Were I to hear these differences with my software, I and every other more or less competent recording engineer would get to the bottom of things, and/or junk the gear and start over. Life is too short to put up with this BS. This is why many audio engineers use the name "audiophool" to describe audiophiles who bumble about without a clue as to what they are doing and then reach totally unfounded conclusions.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Edits: 01/05/12
Incomplete knowledge of what is being done is dangerous. There are so many who think in true or false terms (mostly flase) w/o clarity of what is in the box that I wonder how far we have progressed (or gone back) in the so called 'knowledge society' that so many IT guys brag about.
...what is your take on this "header" difference stuff?
I thought that was the crux of his post.
And it's a big challenge to the bits are bits claim for static FILES.
Two "bit identical" files with different headers sound DIFFERENT.
More to files than just bits?
Fascinating subject. I am *thinking* no but have zero experiments to back up my claim. So, in fairness, I'm not SAYING no!!
Cheers,
Presto
I wasn't just responding to TL's specific point but to the 'grey gloo' about what is in the black boxes in the transmission chain. If one has done proper scientific analysis and programming, one ensures that what software and hardware do in each stage is clear and explicit.
With audio PC playback, there is no such clarity, if only because of IP and of the software 'patching' culture that exixts.
md5sum is not a good tool for checking audio files with headers. There are several wavdiff tools freely available, or sox with proper params can do the job too.
Between the wav tools and MD5 one can see whether the headers or the audio data have changed. These are different changes and they need different tools. I have found that using an editor and mixing the two files out of phase is the best way to compare to files for audio data, in that you can actually hear (and see) the differences. This can tell you a lot about what has changed and possibly why. (For example, a rip of a damaged disk that failed to be secure will often have a single sample that is off and one can hear this as a click at one point in the file.)
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Tony,
I did just that. See my post below about inverting and combining files. I could find absolutely no detectable difference after 20+ generations of converting from wav to flac to wav.
kerkula
You can be sure that the audio samples will not be changed when compressed and expanded. The compression process assures this because of how it works. In addition, there is a double check: the original audio samples are checksummed prior to compression and this is used to verify the decoded samples. The issue, if there is one, will not be in the content of the audio samples in terms of the 0's and 1's that are represented in the file. It will have to be elsewhere, whether in the headers of the file, how the file is stored on disk, etc....
You have to keep in mind that there are people on this forum who believe that two WAV files that are identical also can sound different and that when this happens it is somehow the "fault" of the file and not the equipment used to playback the file. These are the same people who will tell you that "everything matters". Sure, like the color of insulation on the Ethernet cables connecting HDtracks.com's servers to their routers, somehow is going to affect the sound quality that their customers get...
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
I did not read the article either and I'm certain that front end programs are capable of altering header information of files being encoded/decoded. That said, the command-line FLAC encoder/decoder will not alter the md5sum of a file with a round trip of compression/decompression.
$ md5sum "(01)Donna.wav"
1edd5e7e92dbbca6413b1bfa2107fea8 *(01)Donna.wav
$ flac "(01)Donna.wav"
flac 1.2.1, Copyright (C) 2000,2001,2002,2003,2004,2005,2006,2007 Josh Coalson
flac comes with ABSOLUTELY NO WARRANTY. This is free software, and you are
welcome to redistribute it under certain conditions. Type `flac' for details.
(01)Donna.wav: wrote 18105303 bytes, ratio=0.588
$ flac -f -d "(01)Donna.flac"
flac 1.2.1, Copyright (C) 2000,2001,2002,2003,2004,2005,2006,2007 Josh Coalson
flac comes with ABSOLUTELY NO WARRANTY. This is free software, and you are
welcome to redistribute it under certain conditions. Type `flac' for details.
(01)Donna.flac: done
$ md5sum "(01)Donna.wav"
1edd5e7e92dbbca6413b1bfa2107fea8 *(01)Donna.wav
As you noted, some of the advanced command-line options are capable of changing the encoding and headers, but I have not fully experimented with their effect.
And when you compress a data or program file it will no longer work.
Here's a test: Convert WAV to FLAC, back to WAV. Listen for the, ahem, sound degradations. Then repeat the process, each time using the "degraded" new WAV to convert to FLAC. then back to even more degraded WAV. After ten or twenty times, the sound quality should be such that you will be cutting your ears off, at a minimum, or jumping off a building at (perhaps) worst. And keep reading Absolute Sound for advice. ;-)
So I just tried that. I took a 16/44.1 wav file that I created myself playing the guitar. I converted that file to flac and back to wav and back to flac over 20 times.
I listened to the original and then the multi generation file and could hear no difference. But maybe my hearing's not so good or maybe my system's not so good. So I opened both files in Cool edit. I copied the multi generation version and inverted it then mixed it back with the original - essentially subtracting one from the other. I figure there would have to be some residual noise if the two files weren't identical. But low and behold there was no residual noise to be detected down to -150 db. Just total silence.
I'm sure someone will punch a hole in this methodology but I can tell you this: for my ears, and my system, and my money, flac files work just fine.
kerkula
I'm now on medication after finding out that all of my Hi-Res titles purchased from HDtracks, Linn, etc. are corrupted.
But I do enjoy articles like this. I'll take one of my HRx WAV files, convert it to FLAC and back to WAV to see if I can hear a difference.
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 ]
