I actually did, quite a while ago though. Possibly in a periodical before the internet was so mainstream.
Anyway, what I read was that the high end sounds reverberate off of the walls, floor and ceiling and eventually drift down into the range of the human ear. This 'trailing sound', for lack of a better term, is lost on a CD and has a distinguishable effect on the music.
I'm not arguing about the importance of the high frequency components of an audio signal. I'm talking about what's left of the over-20 KHz component of a signal after that signal has been transferred to an LP. LPs might have traces of, say 30 KHz energy, but even if we imagine that such high frequencies add to our perception of music, consider that we're talking about something that in its original form might have been 40, 50, 60 dB or more below the average signal level, then lopping off another 20, 30, 40 dB because of the roll-off imposed by the vinyl?
That's precious little to imagine making so big a difference in the sound of LPs, whereas the amplitudes of euphonic distortions are right up front and in your face, making them much better suspects for why LPs might sound "better" to some people.
Noise is added to CDs too. In fact, these mp3 and AAC files you are listening to apply advanced psychoacoustical tricks to fool you into thinking it's "more real" as well. With the CD format you loose information twice. Once when it is recorded digitally (hopefully 24-bit) and again when it is dithered down to 16-bit for mastering to a CD. There are pros and cons to every format. My point is that the CD is not the end all be all of sound quality.
Noise, called dither, is added to CDs (and digital media) to decorrelate quantization noise. This noise ideally has an amplitude of half of the amplitude that corresponds to the lowest-order sampling bit. Dither slightly raises the noise floor, but spreads the noise out over the entire audio spectrum making it far less perceptible than signal-correlated noise clustering around distinct frequencies.
You might have a point about CDs not being "the end all be all of sound quality", but the significance you seem to attach to "loss of information" doesn't has much to do with it.
Noise is added to CDs too. In fact, these mp3 and AAC files you are listening to apply advanced psychoacoustical tricks to fool you into thinking it's "more real" as well. With the CD format you loose information twice. Once when it is recorded digitally (hopefully 24-bit) and again when it is dithered down to 16-bit for mastering to a CD. There are pros and cons to every format. My point is that the CD is not the end all be all of sound quality.
The CD format (16bit 44.1k sample rate) was decent enough for when it first came out, but it is really time to leave it behind. When you do a/b comparisons of an original 24bit/96k recording and the 16 bit CD version you can really hear the difference. Of course, since we can't get the music industry to agree on a single replacement format, it is taking forever to get a real alternative to the CD.
If the record companies worried half as much about the sound quality as they do about watermarking and copy protection, CDs would have gone the way of 8-track tapes years ago.
You might have a point about CDs not being "the end all be all of sound quality", but the significance you seem to attach to "loss of information" doesn't has much to do with it.
You do lose a lot when going from 24bit/96k to 16bit/44.1k though. See the post above. I suppose there should be only one CD Player on the market because it all sounds the same? Have you ever heard a 1/2inch analog master? Way better than a CD and on par with many 24bit digital recordings. In my opinion this is the point at which digital begins to surpass analog. From the day CDs came on the market I thought they were garbage. I liked the dynamic range and removal of noise of course but they otherwise sounded very two dimensional. Sharp but with no real presence. If it is some psycho acoustical trick that makes analog recordings sound better what does it matter? I am a man not a machine. If it sounds better it sounds better. Not that I believe that is the only factor at work. 24bit/96k is the only digital medium I have heard that sounds good to me. In the meantime I have an iPod with music encoded at various levels of quality and yes, I can hear the difference.
You do lose a lot when going from 24bit/96k to 16bit/44.1k though. See the post above.
First, how much is "a lot" compared to the limits of human hearing? I'd be curious to know how well people do in double blind comparisons of 16/48 with 24/96 -- not self-administered tests when someone is flipping a switch between known choices.
Second, you say this almost as if you don't think there are losses in analog, loss in tranferring from mic to original multi-track tape, tape to other tape during mix down, tape to cutting head, cutting head to stamping master, stamping master to production master, production master to vinyl disc, vinyl disc to needle, etc. Loss upon loss every step of the way. Even rare "direct-to-disc" audiophile LPs only cut out a few of those lossy stages.
Quote:
I suppose there should be only one CD Player on the market because it all sounds the same?
Implementation matters -- you can made CD players, and digital recording, equipment badly. But current technology is good enough that diminishing returns kick in pretty fast, and I do believe that a lot of the self-proclaimed golden-ear crowd are fooling themselves that $5000 CD transports hooked up to $5000 dedicated DACs via $500 specially braided "time aligned" oxygen free digital coax with rhodium plated connectors gain themselves much of anything beyond bragging rights.
Quote:
Have you ever heard a 1/2inch analog master? Way better than a CD and on par with many 24bit digital recordings.
Can't say that I have heard a 1/2 inch analog master. Have you ever heard a 1/2 analog master side by side with a CD of the same, in a level-matched double-blind comparison?
Quote:
In my opinion this is the point at which digital begins to surpass analog. From the day CDs came on the market I thought they were garbage. I liked the dynamic range and removal of noise of course but they otherwise sounded very two dimensional. Sharp but with no real presence. If it is some psycho acoustical trick that makes analog recordings sound better what does it matter?
What does it matter, indeed.
What matters is that very few analog fans will actually admit that noise and distortion could actually be plusses. They typically couch their stated preference for analog in such a way as to claim technical superiority for analog -- more accurate, more information, more resolution. When asked what was missing from CD, you said it was THE MUSIC. That doesn't sound like someone admitting to the possibility that what's really missing is some pleasing type of distortion that merely lends an illusion of greated musicality.
I wish I knew of a more formal test of this than this test I read about in rec.audio.opionion many years ago, but I suspect this one guy's results would be easy to repeat:
If the evil of 16/44.1 or 16/48 digital sampling and quantization truly sucked some musical spark out of recordings that analog LPs preserve, then you'd have to imagine that a digital recording of an LP would be the worst of both worlds: the surface noise, pops, clicks, wow, flutter, poor channel separation, etc., of vinyl PLUS the cold soulessness of mere CD-quality digital sound.
Guess what? People seem to have a hard time telling a DAT tape of an LP from the LP itself. DAT and CD, by the way, use essentially the same digital data format, although the slightly higher 48K sampling rate is often available for DAT over 44.1K -- I can't tell from this article which the guy used. At any rate, far, far short of 24/96 digital. What ever special music quality LPs supposedly possess, it seems well within the realm of basic digital recording to capture that elusive quality.
Bottom line: I can tell the difference between AIFF and 256K AAC on my iPod even with my ears.
Not everyone is the same. Everyone is different. You can't just say that because you don't hear a difference no one can and anyone who disagrees with you is wrong.
Actually, you've failed to prove that. If you can't double blind it to a statistically significant degree then you are literally fooling yourself.
I make no claims about anyone else's hearing unless based on test results they report themselves. I can comment generally on people's hearing because there have been scientifically sound tests done on large numbers of people:
So on the one hand you have a range of audio engineers and classically trained musicians, of a wide variety of ages, listening in a specialised listening room with studio quality gear under scientifically sound conditions, and on the other you have certain people in this thread.
One group finds it difficult to distinguish many tracks at all and finds the difference not unpleasant when distinguishable, the other thinks that you would have to be deaf to miss the glaring lack of quality in compressed music. Hmmm
To finish with a random bit of CD trivia: many early CDs sounded tinny because they used the same master as they did for the LP. Unfortunately they had to boost the high-frequency for the LP to ensure that it came through at acceptable levels. One good reason to buy the remaster and be suspicious of any CD/SACD comparison that doesn't use the same master.
If the record companies worried half as much about the sound quality as they do about watermarking and copy protection, CDs would have gone the way of 8-track tapes years ago.
They love format wars too don't they? SACD, DVD-A and now DualDisc.
Actually, you've failed to prove that. If you can't double blind it to a statistically significant degree then you are literally fooling yourself.
I make no claims about anyone else's hearing unless based on test results they report themselves. I can comment generally on people's hearing because there have been scientifically sound tests done on large numbers of people:
So on the one hand you have a range of audio engineers and classically trained musicians, of a wide variety of ages, listening in a specialised listening room with studio quality gear under scientifically sound conditions, and on the other you have certain people in this thread.
One group finds it difficult to distinguish many tracks at all and finds the difference not unpleasant when distinguishable, the other thinks that you would have to be deaf to miss the glaring lack of quality in compressed music. Hmmm
-snip-
You should read through the full report, the summery is can be a bit missleading.
Here is a synopsis:
The purpose of these tests were to compare MPEG-1 Layer II and Layer III coders vs AAC
There were 31 subjects, and each was placed in one of three location in the listening room.
It was statistically obvious that the seating had a direct bearing on whether a person could tell the deference between test signals. So they had to throw out the data on the third position, leaving only 22 subjects.
Since there was also a statistical difference between the first two positions and the ability to perceive a deference between test signals, they debated if they should use both of the remaining positions or not. They decided that 11 subjects were not really enough, and since the primary purpose of the test was to compare the coders, which remained constant in both positions, they they pooled the data for seats 1 and 2.
Figure 5 in the document shows that while ACC is statistically distinguishable from the original signal, it is superior to MPEG-1 Layer II and Layer III.
-----
Since no one on this board has said that they can tell the difference between the original recordings of castanets, pitch pipes, human speech, etc., from ACC, the above test is useless (even though it does show that people can differentiate ACC and the original signal).
I'd bet money that if you took the same subjects of the test, gave them a good set of headphones and used a bunch music CDs instead of the signals used in the above test, that they would be able to distinguish the difference between original songs and ACC encoded at least 8 out of 10 times.
Hearing is believing: try a/b comparing some ACC vs original songs yourself, most likely you will be able to tell the difference.
You should read through the full report, the summery is can be a bit missleading.
I have read it and yes, your summary is misleading.
* Castanets and harpsichords are very difficult to psycho-acoustically encode.
* Suzanne Vega, Tracy Chapman, Dire Straits, Ornette Coleman are all popular musicians.
The results, as I said, show that 128kbps AAC is statistically distinguishable from the source, but not unpleasantly so and not on all samples.
Notice: 128kbps (done using a 1996-era encoder as well)
You keep telling people to a/b test, which leads me to suppose that you don't understand why double-blind abx tests are important. Can you explain the 'x' and why it's needed to sensibly test 'a' vs. 'b'? If so, then why tell people to a/b test?
Also, if you'd read my earlier post you'd know that getting it right 8 times out of 10 isn't statistically significant. So you've just confidently predicted that people listening to music on headphones would *fail* to distinguish them to a statistically significant degree, which is *less* than the report claims for speaker listening.
Hearing is *not* believing, and folk wisdom or common sense does not trump the scientific method in this case.
Hearing is believing: try a/b comparing some ACC vs original songs yourself, most likely you will be able to tell the difference.
Apparently that's not good enough for some. Like Tonton, who will flat out call you a liar if you dare to do such an outlandish test. Yes, I've done blind tests (not double blind) and was able to discern between AAC 192 and AIFF.
Human hearing is very complicated and has more to do with brain function than anything else. The mere existence of the phenomenon known as "perfect pitch" should be enough to convince us that some people can perceive much more than others. Just how is it that someone can know the exact frequency of a waveform to such precision? People with perfect pitch can name which note you're playing, and even whether it is slightly sharp or flat. That's like being able to name and discern between a 12000 and 12100 Hz tone. Perfect pitch is a well-documented phenomenon.
Rest assured, the difference between AAC and CD sound is immediately obvious to some people. Though, I've yet to decide whether this capability is a blessing or annoyance.
Rest assured, the difference between AAC and CD sound is immediately obvious to some people. Though, I've yet to decide whether this capability is a blessing or annoyance.
Unfortunately the difference between two *identical* signals is also immediately obvious to some people, as long as they are told that the signal has cheap cables/digital encoding/green ink on the CD/bad feng shui etc.
This capability is, without doubt, an annoyance, so anyone claiming to clearly hear differences that are at best subtle to everyone else should at least demonstrate that they understand and account for the placebo effect, or they're basically wasting everyone's time.
Unfortunately the difference between two *identical* signals is also immediately obvious to some people, as long as they are told that the signal has cheap cables/digital encoding/green ink on the CD/bad feng shui etc.
I've set up some blind-as-I-can-make-them tests for myself using shuffle play, and I know I can hear some really obvious things like hard-to-encode sample tracks that others have told me about, and my own tests of things like two-pass 128K AAC (meaning that I took a CD track, encoded it to ACC at 128K, decoded it, then re-encoded the decoded version -- nasty!). I think the more you listen to worst-case scenarios, the more you can train yourself to notice less obvious distortions.
Whether you're really doing yourself a favor by training your ears like this is another matter.
I would prefer it if Apple made 192K downloads available, but I'm still reasonably happy with 128K. While I don't doubt the ability of many people to distinguish differences between 128K AAC and original CD tracks, I do think that many criticisms about AAC as being "absolutely awful", "dull and lifeless", "worthless shit", etc. are perceptual exagerrations driven more by suspicion of and prejudice against compression technology than by the actual quality of what's being heard.
This capability is, without doubt, an annoyance, so anyone claiming to clearly hear differences that are at best subtle to everyone else should at least demonstrate that they understand and account for the placebo effect, or they're basically wasting everyone's time.
Ah. I see where you're coming from. You're annoyed, plain and simple. The placebo effect is well documented.
Yet, please don't take out your annoyance on those of us who can actually tell the difference in sonic fidelity.
Yet, please don't take out your annoyance on those of us who can actually tell the difference in sonic fidelity.
OK, and how do I tell which is which?
My general rules of thumb (not guaranteed by any means) to spot the people talking nonsense are:
* anyone who mentions that they're an audio engineer or musician as if that exempts them from the laws of physics or the placebo effect.
* anyone who talks about how expensive their equipment is as if poverty was the only thing keeping people happy with AAC/MP3
* anyone who shows a general lack of understanding about abx testing / psycho-acoustic encoding / digital audio / physics / science / statistics but is happy to make proclamations about things in those areas.
* anyone who acts like a total prima donna about compressed music and how it hurts for their ears to suffer it or starts making wild accusations about AAC, MP3, compressed music, or digital music in general as if nothing but the soundwaves direct from the artist's instrument (or LPs \) will ever be enough for them.
* people who claim to have done tests yet provide no further details than "I can tell xkbps from source" as if that statement actually tells us anything e.g. which samples? which encoder? blind? double-blind? equipment? methodology?
* Anyone who makes claims based on their own alledgedly golden ears without qualifying that, even if they are correct, they realize that they are an anomaly, probably more cursed than blessed, as you say.
Make an honest effort to distinguish yourself from these people and you will be in no danger of annoying me. And if you *are* one of these people then you're still not annoying me, you're just wrong.
edit: and just to be clear, I've abx'd AAC samples against the source myself, in fact with certain combinations of bitrate, encoder and audio sample it's hard not to. So no-one is claiming that AAC is always 'CD quality' (whatever that means). But that doesn't translate to all digital audio, all AAC encoded music or even all ACC encoded music at a given bitrate sounding horrible. It's a numbers game and overall AAC does fine, even at 128kbps.
It seems as if you've made a sport of hunting down 'self-deceivers', as if identifying and ridiculing these people is paramount when discussing audio compression.
Yes, they certainly exhist. But when you start with the assumption and proclomation that everyone is self-deceived, you're unlikely to have a healthy discussion. Hopefully, you don't handle all topics in a similar manner.
Yes, they certainly exist. But when you start with the assumption and proclomation that everyone is self-deceived, you're unlikely to have a healthy discussion. Hopefully, you don't handle all topics in a similar manner.
No, that is exactly the point. Everyone *is* self-deceived, all the time, about almost everything they perceive. If you don't accept that basic fact then you can't get anywhere when discussing *psycho*-acoustic compression (which even disregarding the placebo effect *relies* on fooling your brain/ears to compress audio). So if you don't accept self-deception then you might as well give up as there is no longer a rational basis on which to continue a "healthy" discussion.
I've been on both sides of this argument. I believe that buying speaker cables with a comma in the price is perfectly justifiable yet most of the time I listen to cheap car radios or my iPod with AAC 192K.
This debate reminds me of the protagonist in Herman Hesse's Stepenwolf who could not enjoy the Mozart coming off the victrola LP due to the sound while his new friend could only hear the music and not the noise. Who was better off?
I can empathise with the view that most listening is affected by factors which are subjective and hidden from our own awareness.
Having said that, it is puerile to dismiss the fact that lower encoding is inaudible.
Mark Levinson, an audio dealer had a curtain erected in his store and behind the curtain he had a piano and a 24Bit recording setup and a very high quality audio system. On occasion he would invite musicians and others to try to guess which they were hearing... live or recording.... Best results were often at chance level...Try that test with you 128K recordings.. I'd bet a dime which way the results would turn out.
I have read it and yes, your summary is misleading.
* Castanets and harpsichords are very difficult to psycho-acoustically encode.
* Suzanne Vega, Tracy Chapman, Dire Straits, Ornette Coleman are all popular musicians.
The results, as I said, show that 128kbps AAC is statistically distinguishable from the source, but not unpleasantly so and not on all samples.
Notice: 128kbps (done using a 1996-era encoder as well)
Right. The results say that it is possible to tell the difference between the encoded and original signal, but for some illogical reason you go against the study and say that anyone who claims that they can hear the difference is being fooled by the placebo effect.
Truthfully the "Report on the MPEG-2 AAC Stereo Verification Tests" is meaningless to our debate. It might be of some use if they would give us the raw data, but even then the listening conditions were far from optimal (the subjects were not allowed to adjust their positions or the volume of the signals) and 22 people is nowhere near a lage enough sample to reach any conclusions.
Quote:
You keep telling people to a/b test, which leads me to suppose that you don't understand why double-blind abx tests are important. Can you explain the 'x' and why it's needed to sensibly test 'a' vs. 'b'? If so, then why tell people to a/b test?
I fully understand the the reasoning and necessity for double-blind testing. I have also noticed that the vast majority of people who claim that there is no discernible difference between encoded formats vs the original are "talking out their ass" and have have never objectively listened to various songs in a/b comparisons. You don't need a double-blind test to determine if you can taste a difference between coke and Pepsi, there is an obvious difference in taste. With many songs the effects of compression are just as obvious.
For the people who want to do a more scientific comparison, you can do so using iTunes. Just rip a song in AIFF and AAC and put them in a playlist together. Then select the playlist and turn on shuffle play. Close your eyes and hit the play button.
Quote:
Also, if you'd read my earlier post you'd know that getting it right 8 times out of 10 isn't statistically significant. So you've just confidently predicted that people listening to music on headphones would *fail* to distinguish them to a statistically significant degree, which is *less* than the report claims for speaker listening.
"Getting it right eight out of ten times" was meant as a ratio, you would have to listen to far more samples then ten to get proper data. I was saying that the average person should be able hear the difference between original and AAC encoded songs with greater then 80% accuracy (it really depends on the song in question -- on some songs the encoding is quite obvious, while others are nearly indistinguishable from the original ).
Quote:
Hearing is *not* believing, and folk wisdom or common sense does not trump the scientific method in this case.
Next you'll be claiming "the camera never lies".
Actually, you are the one who is going against the the scientific method. Doing a quick search I have found nothing on-line that backs up your belief that any differences heard are only due to placebo effects, and the one document that you introduced shows that it is possible for people to distinguish between AAC encoded material and the original.
I really don't understand your irrational belief that people cannot distinguish between encoded music and the original songs. Maybe you can't hear a difference, but others can (even the study you introduced says so).
I've set up some blind-as-I-can-make-them tests for myself using shuffle play, and I know I can hear some really obvious things like hard-to-encode sample tracks that others have told me about, and my own tests of things like two-pass 128K AAC (meaning that I took a CD track, encoded it to ACC at 128K, decoded it, then re-encoded the decoded version -- nasty!). I think the more you listen to worst-case scenarios, the more you can train yourself to notice less obvious distortions.
Whether you're really doing yourself a favor by training your ears like this is another matter.
I would prefer it if Apple made 192K downloads available, but I'm still reasonably happy with 128K. While I don't doubt the ability of many people to distinguish differences between 128K AAC and original CD tracks, I do think that many criticisms about AAC as being "absolutely awful", "dull and lifeless", "worthless shit", etc. are perceptual exagerrations driven more by suspicion of and prejudice against compression technology than by the actual quality of what's being heard.
Agreed. 128K AAC is better then cassettes that people used to buy in droves for there portability, and it is not nearly as bad as some people claim. That being said, I would buy more form the iTunes music store if they gave me access to higher bit rates. As it is I usually just by the CD and rip it myself.
Comments
Originally posted by bunge
I actually did, quite a while ago though. Possibly in a periodical before the internet was so mainstream.
Anyway, what I read was that the high end sounds reverberate off of the walls, floor and ceiling and eventually drift down into the range of the human ear. This 'trailing sound', for lack of a better term, is lost on a CD and has a distinguishable effect on the music.
I'm not arguing about the importance of the high frequency components of an audio signal. I'm talking about what's left of the over-20 KHz component of a signal after that signal has been transferred to an LP. LPs might have traces of, say 30 KHz energy, but even if we imagine that such high frequencies add to our perception of music, consider that we're talking about something that in its original form might have been 40, 50, 60 dB or more below the average signal level, then lopping off another 20, 30, 40 dB because of the roll-off imposed by the vinyl?
That's precious little to imagine making so big a difference in the sound of LPs, whereas the amplitudes of euphonic distortions are right up front and in your face, making them much better suspects for why LPs might sound "better" to some people.
Originally posted by TWinbrook46636
Noise is added to CDs too. In fact, these mp3 and AAC files you are listening to apply advanced psychoacoustical tricks to fool you into thinking it's "more real" as well. With the CD format you loose information twice. Once when it is recorded digitally (hopefully 24-bit) and again when it is dithered down to 16-bit for mastering to a CD. There are pros and cons to every format. My point is that the CD is not the end all be all of sound quality.
Noise, called dither, is added to CDs (and digital media) to decorrelate quantization noise. This noise ideally has an amplitude of half of the amplitude that corresponds to the lowest-order sampling bit. Dither slightly raises the noise floor, but spreads the noise out over the entire audio spectrum making it far less perceptible than signal-correlated noise clustering around distinct frequencies.
You might have a point about CDs not being "the end all be all of sound quality", but the significance you seem to attach to "loss of information" doesn't has much to do with it.
Originally posted by TWinbrook46636
Noise is added to CDs too. In fact, these mp3 and AAC files you are listening to apply advanced psychoacoustical tricks to fool you into thinking it's "more real" as well. With the CD format you loose information twice. Once when it is recorded digitally (hopefully 24-bit) and again when it is dithered down to 16-bit for mastering to a CD. There are pros and cons to every format. My point is that the CD is not the end all be all of sound quality.
The CD format (16bit 44.1k sample rate) was decent enough for when it first came out, but it is really time to leave it behind. When you do a/b comparisons of an original 24bit/96k recording and the 16 bit CD version you can really hear the difference. Of course, since we can't get the music industry to agree on a single replacement format, it is taking forever to get a real alternative to the CD.
If the record companies worried half as much about the sound quality as they do about watermarking and copy protection, CDs would have gone the way of 8-track tapes years ago.
Originally posted by shetline
You might have a point about CDs not being "the end all be all of sound quality", but the significance you seem to attach to "loss of information" doesn't has much to do with it.
You do lose a lot when going from 24bit/96k to 16bit/44.1k though. See the post above. I suppose there should be only one CD Player on the market because it all sounds the same? Have you ever heard a 1/2inch analog master? Way better than a CD and on par with many 24bit digital recordings. In my opinion this is the point at which digital begins to surpass analog. From the day CDs came on the market I thought they were garbage. I liked the dynamic range and removal of noise of course but they otherwise sounded very two dimensional. Sharp but with no real presence. If it is some psycho acoustical trick that makes analog recordings sound better what does it matter? I am a man not a machine. If it sounds better it sounds better. Not that I believe that is the only factor at work. 24bit/96k is the only digital medium I have heard that sounds good to me. In the meantime I have an iPod with music encoded at various levels of quality and yes, I can hear the difference.
Originally posted by shetline
digital_signal - original_signal = noise
Do a little simple algebra...
digital_signal = original_signal + noise
I think it is more like this:
analog_signal = original_signal + noise
-vs-
digital_signal = original_signal_facsimile - noise
Originally posted by TWinbrook46636
You do lose a lot when going from 24bit/96k to 16bit/44.1k though. See the post above.
First, how much is "a lot" compared to the limits of human hearing? I'd be curious to know how well people do in double blind comparisons of 16/48 with 24/96 -- not self-administered tests when someone is flipping a switch between known choices.
Second, you say this almost as if you don't think there are losses in analog, loss in tranferring from mic to original multi-track tape, tape to other tape during mix down, tape to cutting head, cutting head to stamping master, stamping master to production master, production master to vinyl disc, vinyl disc to needle, etc. Loss upon loss every step of the way. Even rare "direct-to-disc" audiophile LPs only cut out a few of those lossy stages.
I suppose there should be only one CD Player on the market because it all sounds the same?
Implementation matters -- you can made CD players, and digital recording, equipment badly. But current technology is good enough that diminishing returns kick in pretty fast, and I do believe that a lot of the self-proclaimed golden-ear crowd are fooling themselves that $5000 CD transports hooked up to $5000 dedicated DACs via $500 specially braided "time aligned" oxygen free digital coax with rhodium plated connectors gain themselves much of anything beyond bragging rights.
Have you ever heard a 1/2inch analog master? Way better than a CD and on par with many 24bit digital recordings.
Can't say that I have heard a 1/2 inch analog master. Have you ever heard a 1/2 analog master side by side with a CD of the same, in a level-matched double-blind comparison?
In my opinion this is the point at which digital begins to surpass analog. From the day CDs came on the market I thought they were garbage. I liked the dynamic range and removal of noise of course but they otherwise sounded very two dimensional. Sharp but with no real presence. If it is some psycho acoustical trick that makes analog recordings sound better what does it matter?
What does it matter, indeed.
What matters is that very few analog fans will actually admit that noise and distortion could actually be plusses. They typically couch their stated preference for analog in such a way as to claim technical superiority for analog -- more accurate, more information, more resolution. When asked what was missing from CD, you said it was THE MUSIC. That doesn't sound like someone admitting to the possibility that what's really missing is some pleasing type of distortion that merely lends an illusion of greated musicality.
I wish I knew of a more formal test of this than this test I read about in rec.audio.opionion many years ago, but I suspect this one guy's results would be easy to repeat:
A comparison between vinyl and a digital recording of the same vinyl (The point of interest is about half way through this text.)
If the evil of 16/44.1 or 16/48 digital sampling and quantization truly sucked some musical spark out of recordings that analog LPs preserve, then you'd have to imagine that a digital recording of an LP would be the worst of both worlds: the surface noise, pops, clicks, wow, flutter, poor channel separation, etc., of vinyl PLUS the cold soulessness of mere CD-quality digital sound.
Guess what? People seem to have a hard time telling a DAT tape of an LP from the LP itself. DAT and CD, by the way, use essentially the same digital data format, although the slightly higher 48K sampling rate is often available for DAT over 44.1K -- I can't tell from this article which the guy used. At any rate, far, far short of 24/96 digital. What ever special music quality LPs supposedly possess, it seems well within the realm of basic digital recording to capture that elusive quality.
Originally posted by TWinbrook46636
Bottom line: I can tell the difference between AIFF and 256K AAC on my iPod even with my ears.
Not everyone is the same. Everyone is different. You can't just say that because you don't hear a difference no one can and anyone who disagrees with you is wrong.
Actually, you've failed to prove that. If you can't double blind it to a statistically significant degree then you are literally fooling yourself.
I make no claims about anyone else's hearing unless based on test results they report themselves. I can comment generally on people's hearing because there have been scientifically sound tests done on large numbers of people:
summary report
full report PDF
So on the one hand you have a range of audio engineers and classically trained musicians, of a wide variety of ages, listening in a specialised listening room with studio quality gear under scientifically sound conditions, and on the other you have certain people in this thread.
One group finds it difficult to distinguish many tracks at all and finds the difference not unpleasant when distinguishable, the other thinks that you would have to be deaf to miss the glaring lack of quality in compressed music. Hmmm
Here's a band you might like: http://phobos.apple.com/WebObjects/M...rtistId=649817
To finish with a random bit of CD trivia: many early CDs sounded tinny because they used the same master as they did for the LP. Unfortunately they had to boost the high-frequency for the LP to ensure that it came through at acceptable levels. One good reason to buy the remaster and be suspicious of any CD/SACD comparison that doesn't use the same master.
Originally posted by Res
If the record companies worried half as much about the sound quality as they do about watermarking and copy protection, CDs would have gone the way of 8-track tapes years ago.
They love format wars too don't they? SACD, DVD-A and now DualDisc.
Originally posted by stupider...likeafox
Actually, you've failed to prove that. If you can't double blind it to a statistically significant degree then you are literally fooling yourself.
I make no claims about anyone else's hearing unless based on test results they report themselves. I can comment generally on people's hearing because there have been scientifically sound tests done on large numbers of people:
summary report
full report PDF
So on the one hand you have a range of audio engineers and classically trained musicians, of a wide variety of ages, listening in a specialised listening room with studio quality gear under scientifically sound conditions, and on the other you have certain people in this thread.
One group finds it difficult to distinguish many tracks at all and finds the difference not unpleasant when distinguishable, the other thinks that you would have to be deaf to miss the glaring lack of quality in compressed music. Hmmm
-snip-
You should read through the full report, the summery is can be a bit missleading.
Here is a synopsis:
The purpose of these tests were to compare MPEG-1 Layer II and Layer III coders vs AAC
There were 31 subjects, and each was placed in one of three location in the listening room.
It was statistically obvious that the seating had a direct bearing on whether a person could tell the deference between test signals. So they had to throw out the data on the third position, leaving only 22 subjects.
Since there was also a statistical difference between the first two positions and the ability to perceive a deference between test signals, they debated if they should use both of the remaining positions or not. They decided that 11 subjects were not really enough, and since the primary purpose of the test was to compare the coders, which remained constant in both positions, they they pooled the data for seats 1 and 2.
The test signals used:
Signal (duration in seconds)
Castanets (16.6)
Harpsichord (18.2)
Pitch Pipe (28.4)
Glockenspiel (22.7)
Male German Speech (20.5)
Suzanne Vega, Tom?s Diner Solitude Standing (21.0)
Tracy Chapman (29.5)
Ornette Coleman Dreams (22.9)
Accordion/Triangle Private (analogue) recording (19.9)
Dire Straits (33.7)
Conclusions of the test:
Figure 5 in the document shows that while ACC is statistically distinguishable from the original signal, it is superior to MPEG-1 Layer II and Layer III.
-----
Since no one on this board has said that they can tell the difference between the original recordings of castanets, pitch pipes, human speech, etc., from ACC, the above test is useless (even though it does show that people can differentiate ACC and the original signal).
I'd bet money that if you took the same subjects of the test, gave them a good set of headphones and used a bunch music CDs instead of the signals used in the above test, that they would be able to distinguish the difference between original songs and ACC encoded at least 8 out of 10 times.
Hearing is believing: try a/b comparing some ACC vs original songs yourself, most likely you will be able to tell the difference.
Originally posted by Res
You should read through the full report, the summery is can be a bit missleading.
I have read it and yes, your summary is misleading.
* Castanets and harpsichords are very difficult to psycho-acoustically encode.
* Suzanne Vega, Tracy Chapman, Dire Straits, Ornette Coleman are all popular musicians.
The results, as I said, show that 128kbps AAC is statistically distinguishable from the source, but not unpleasantly so and not on all samples.
Notice: 128kbps (done using a 1996-era encoder as well)
You keep telling people to a/b test, which leads me to suppose that you don't understand why double-blind abx tests are important. Can you explain the 'x' and why it's needed to sensibly test 'a' vs. 'b'? If so, then why tell people to a/b test?
Also, if you'd read my earlier post you'd know that getting it right 8 times out of 10 isn't statistically significant. So you've just confidently predicted that people listening to music on headphones would *fail* to distinguish them to a statistically significant degree, which is *less* than the report claims for speaker listening.
Hearing is *not* believing, and folk wisdom or common sense does not trump the scientific method in this case.
Next you'll be claiming "the camera never lies".
Originally posted by Res
Hearing is believing: try a/b comparing some ACC vs original songs yourself, most likely you will be able to tell the difference.
Apparently that's not good enough for some.
Human hearing is very complicated and has more to do with brain function than anything else. The mere existence of the phenomenon known as "perfect pitch" should be enough to convince us that some people can perceive much more than others. Just how is it that someone can know the exact frequency of a waveform to such precision? People with perfect pitch can name which note you're playing, and even whether it is slightly sharp or flat. That's like being able to name and discern between a 12000 and 12100 Hz tone. Perfect pitch is a well-documented phenomenon.
Rest assured, the difference between AAC and CD sound is immediately obvious to some people. Though, I've yet to decide whether this capability is a blessing or annoyance.
Originally posted by dfiler
Rest assured, the difference between AAC and CD sound is immediately obvious to some people. Though, I've yet to decide whether this capability is a blessing or annoyance.
Unfortunately the difference between two *identical* signals is also immediately obvious to some people, as long as they are told that the signal has cheap cables/digital encoding/green ink on the CD/bad feng shui etc.
This capability is, without doubt, an annoyance, so anyone claiming to clearly hear differences that are at best subtle to everyone else should at least demonstrate that they understand and account for the placebo effect, or they're basically wasting everyone's time.
Originally posted by stupider...likeafox
Unfortunately the difference between two *identical* signals is also immediately obvious to some people, as long as they are told that the signal has cheap cables/digital encoding/green ink on the CD/bad feng shui etc.
I've set up some blind-as-I-can-make-them tests for myself using shuffle play, and I know I can hear some really obvious things like hard-to-encode sample tracks that others have told me about, and my own tests of things like two-pass 128K AAC (meaning that I took a CD track, encoded it to ACC at 128K, decoded it, then re-encoded the decoded version -- nasty!). I think the more you listen to worst-case scenarios, the more you can train yourself to notice less obvious distortions.
Whether you're really doing yourself a favor by training your ears like this is another matter.
I would prefer it if Apple made 192K downloads available, but I'm still reasonably happy with 128K. While I don't doubt the ability of many people to distinguish differences between 128K AAC and original CD tracks, I do think that many criticisms about AAC as being "absolutely awful", "dull and lifeless", "worthless shit", etc. are perceptual exagerrations driven more by suspicion of and prejudice against compression technology than by the actual quality of what's being heard.
Originally posted by stupider...likeafox
This capability is, without doubt, an annoyance, so anyone claiming to clearly hear differences that are at best subtle to everyone else should at least demonstrate that they understand and account for the placebo effect, or they're basically wasting everyone's time.
Ah. I see where you're coming from. You're annoyed, plain and simple. The placebo effect is well documented.
Yet, please don't take out your annoyance on those of us who can actually tell the difference in sonic fidelity.
Originally posted by dfiler
Yet, please don't take out your annoyance on those of us who can actually tell the difference in sonic fidelity.
OK, and how do I tell which is which?
My general rules of thumb (not guaranteed by any means) to spot the people talking nonsense are:
* anyone who mentions that they're an audio engineer or musician as if that exempts them from the laws of physics or the placebo effect.
* anyone who talks about how expensive their equipment is as if poverty was the only thing keeping people happy with AAC/MP3
* anyone who shows a general lack of understanding about abx testing / psycho-acoustic encoding / digital audio / physics / science / statistics but is happy to make proclamations about things in those areas.
* anyone who acts like a total prima donna about compressed music and how it hurts for their ears to suffer it or starts making wild accusations about AAC, MP3, compressed music, or digital music in general as if nothing but the soundwaves direct from the artist's instrument (or LPs
* people who claim to have done tests yet provide no further details than "I can tell xkbps from source" as if that statement actually tells us anything e.g. which samples? which encoder? blind? double-blind? equipment? methodology?
* Anyone who makes claims based on their own alledgedly golden ears without qualifying that, even if they are correct, they realize that they are an anomaly, probably more cursed than blessed, as you say.
Make an honest effort to distinguish yourself from these people and you will be in no danger of annoying me. And if you *are* one of these people then you're still not annoying me, you're just wrong.
edit: and just to be clear, I've abx'd AAC samples against the source myself, in fact with certain combinations of bitrate, encoder and audio sample it's hard not to. So no-one is claiming that AAC is always 'CD quality' (whatever that means). But that doesn't translate to all digital audio, all AAC encoded music or even all ACC encoded music at a given bitrate sounding horrible. It's a numbers game and overall AAC does fine, even at 128kbps.
Yes, they certainly exhist. But when you start with the assumption and proclomation that everyone is self-deceived, you're unlikely to have a healthy discussion. Hopefully, you don't handle all topics in a similar manner.
Originally posted by dfiler
Yes, they certainly exist. But when you start with the assumption and proclomation that everyone is self-deceived, you're unlikely to have a healthy discussion. Hopefully, you don't handle all topics in a similar manner.
No, that is exactly the point. Everyone *is* self-deceived, all the time, about almost everything they perceive. If you don't accept that basic fact then you can't get anywhere when discussing *psycho*-acoustic compression (which even disregarding the placebo effect *relies* on fooling your brain/ears to compress audio). So if you don't accept self-deception then you might as well give up as there is no longer a rational basis on which to continue a "healthy" discussion.
This debate reminds me of the protagonist in Herman Hesse's Stepenwolf who could not enjoy the Mozart coming off the victrola LP due to the sound while his new friend could only hear the music and not the noise. Who was better off?
I can empathise with the view that most listening is affected by factors which are subjective and hidden from our own awareness.
Having said that, it is puerile to dismiss the fact that lower encoding is inaudible.
Mark Levinson, an audio dealer had a curtain erected in his store and behind the curtain he had a piano and a 24Bit recording setup and a very high quality audio system. On occasion he would invite musicians and others to try to guess which they were hearing... live or recording.... Best results were often at chance level...Try that test with you 128K recordings.. I'd bet a dime which way the results would turn out.
Originally posted by stupider...likeafox
I have read it and yes, your summary is misleading.
* Castanets and harpsichords are very difficult to psycho-acoustically encode.
* Suzanne Vega, Tracy Chapman, Dire Straits, Ornette Coleman are all popular musicians.
The results, as I said, show that 128kbps AAC is statistically distinguishable from the source, but not unpleasantly so and not on all samples.
Notice: 128kbps (done using a 1996-era encoder as well)
Right. The results say that it is possible to tell the difference between the encoded and original signal, but for some illogical reason you go against the study and say that anyone who claims that they can hear the difference is being fooled by the placebo effect.
Truthfully the "Report on the MPEG-2 AAC Stereo Verification Tests" is meaningless to our debate. It might be of some use if they would give us the raw data, but even then the listening conditions were far from optimal (the subjects were not allowed to adjust their positions or the volume of the signals) and 22 people is nowhere near a lage enough sample to reach any conclusions.
You keep telling people to a/b test, which leads me to suppose that you don't understand why double-blind abx tests are important. Can you explain the 'x' and why it's needed to sensibly test 'a' vs. 'b'? If so, then why tell people to a/b test?
I fully understand the the reasoning and necessity for double-blind testing. I have also noticed that the vast majority of people who claim that there is no discernible difference between encoded formats vs the original are "talking out their ass" and have have never objectively listened to various songs in a/b comparisons. You don't need a double-blind test to determine if you can taste a difference between coke and Pepsi, there is an obvious difference in taste. With many songs the effects of compression are just as obvious.
For the people who want to do a more scientific comparison, you can do so using iTunes. Just rip a song in AIFF and AAC and put them in a playlist together. Then select the playlist and turn on shuffle play. Close your eyes and hit the play button.
Also, if you'd read my earlier post you'd know that getting it right 8 times out of 10 isn't statistically significant. So you've just confidently predicted that people listening to music on headphones would *fail* to distinguish them to a statistically significant degree, which is *less* than the report claims for speaker listening.
"Getting it right eight out of ten times" was meant as a ratio, you would have to listen to far more samples then ten to get proper data. I was saying that the average person should be able hear the difference between original and AAC encoded songs with greater then 80% accuracy (it really depends on the song in question -- on some songs the encoding is quite obvious, while others are nearly indistinguishable from the original ).
Hearing is *not* believing, and folk wisdom or common sense does not trump the scientific method in this case.
Next you'll be claiming "the camera never lies".
Actually, you are the one who is going against the the scientific method. Doing a quick search I have found nothing on-line that backs up your belief that any differences heard are only due to placebo effects, and the one document that you introduced shows that it is possible for people to distinguish between AAC encoded material and the original.
I really don't understand your irrational belief that people cannot distinguish between encoded music and the original songs. Maybe you can't hear a difference, but others can (even the study you introduced says so).
Originally posted by shetline
I've set up some blind-as-I-can-make-them tests for myself using shuffle play, and I know I can hear some really obvious things like hard-to-encode sample tracks that others have told me about, and my own tests of things like two-pass 128K AAC (meaning that I took a CD track, encoded it to ACC at 128K, decoded it, then re-encoded the decoded version -- nasty!). I think the more you listen to worst-case scenarios, the more you can train yourself to notice less obvious distortions.
Whether you're really doing yourself a favor by training your ears like this is another matter.
I would prefer it if Apple made 192K downloads available, but I'm still reasonably happy with 128K. While I don't doubt the ability of many people to distinguish differences between 128K AAC and original CD tracks, I do think that many criticisms about AAC as being "absolutely awful", "dull and lifeless", "worthless shit", etc. are perceptual exagerrations driven more by suspicion of and prejudice against compression technology than by the actual quality of what's being heard.
Agreed. 128K AAC is better then cassettes that people used to buy in droves for there portability, and it is not nearly as bad as some people claim. That being said, I would buy more form the iTunes music store if they gave me access to higher bit rates. As it is I usually just by the CD and rip it myself.