I'm not trying to be adversarial. I am, however, trying to clear up a matter that is often misunderstood. Your posts demonstrate a misunderstanding of the math at work. A reproduction of a signal can be broken down into a simple equation: reproduction = original signal + noise. Noise is a loose term. It just means "unwanted signal." Vinyl recording equipment is noise prone. It is subject to EM noise, temperature, material variance, gravity variance, and just about everything else. Digital recording isn't.
It's very difficult to reference audio reproductions to the original source, since ultimately some noise-prone delay feature is required to compare the source to the reproduction. However, it's easy to run a spectrogram of a record and a CD. The CD has much less discernible noise. This means quantitatively that the CD is a better, more accurate reproduction of the original signal.
Sammi is also very misled. Nyquist theorem states that a discrete waveform sampled at twice the maximum frequency of a continuous waveform can be converted perfectly into the continuous waveform via an ideal lowpass filter at half the Nyquist rate. So 44.1kHz sampling is sufficient to perfectly reproduce 20kHz tones. Unfortunately, there are no ideal low pass filters, but oversampling DSPs can do a damn-near-ideal job.
The other problem is that CDs are not merely sampled signals. They are quantized as well, so there's sowe rounding error from that process. Nonetheless, this error is less than the error [noise] introduced by analog mastering and recording. That's a mathematic fact.
I understand you're not trying to be adversarial, but you cannot get around a simple fact: Digital recording merely approximates the sound wave, no matter if it's quantized or filtered or both. It's still an estimation. Period.
However, in the real listening world, there are other factors (as you point out)...noise and susceptibility to heat, moisture, dust, warping, etc. Therefore, most people are going to hear a digital recording as being superior. Certainly the durability and convenience of it make up for any shortcomings, at least in my view.
My point is simply that a great many people feel they can hear a certain warmth and space in an excellent vinyl recording (depending on musical material, and depending on the equipment that it's played on) that is not present in a digital recording. Additionally, any digital recording has to pass through a D/A converter, which can further distort the accuracy. Ultimately, you'll hear the sound being pushed through speakers. With digital, it must first be converted. No matter how good the D/A converter is, it can't be 100% efficient.
You can try to "math your way" out it, but my money is on this: If I had you listen to a brand new analog classical recording on hi-fi equipment, then contrasted it with a CD of the same, you'd hear what I'm talking about. It's not enough to make me embrace vinyl, but it is there.
What I am interesting in doing is illustrating that a digitized recording can often be a more accurate reproduction of the original waveform that an can be an analog recording, and that in practice this is almost always the case.
That is blatantly false. It is not as accurate. It may sound better because of the lack of noise, but it's not more accurate, especially by the time it hits your ears.
Quote:
To this extent I think we can all go home happy, because as far as I can tell no one is any longer in disagreement.
Digital is here to stay, so I would hope you're right.
That is blatantly false. It is not as accurate. It may sound better because of the lack of noise, but it's not more accurate, especially by the time it hits your ears.
You are wrong. Period. (as you so like to say).
Both digital and analogue recording to vinyl add noise (quantisation with dithering is exactly the same as adding random noise to the signal, that you clearly do not understand this Splinemodel and I are unable to help), vinyl adds much, much more and the digital signal is therefore more accurate.
Have you ever taken the output of a record deck, and passed it through a 16-bit/44.1 kHz AD-DA loop and compared the two pathways (i.e. "straight from record deck -> amplifier -> speaker" and "record deck->44.1 kHz/16 bit AD converter -> DA converter -> amplifier -> speaker)? I'm willing to bet you that in a properly controlled double-blind experiment you wouldn't be able to tell the difference with 95% confidence or better.
You can't take a commercial vinyl recording and commercial CD recording of allegedly the same material and use that as the comparison. CDs go through a mastering process that sadly in many cases makes a mockery of CD's enhanced dynamic range. Some folk just think "louder=better" and compress the hell out of the original signal's dynamic range (making the whole lot louder) before pressing it to CD.
Quote:
Originally Posted by sammi jo
In my case, I only have a CD or digital collection, so there was no nostalgia element for me. But the difference in listening experience between the CD and the vinyl was quite marked, both for me and the others in the listening room. (But there may have been some nostalgia element as all the other parties had vinyl collections as well as CDs!).
Sounds like the test wasn't double-blind. Sadly this means it had/has no validity whatsoever.
Quote:
Originally Posted by sammi jo
CDs being a severe compromise to start with
CD is not a severe compromise. It is perfectly adequate.
Quote:
Originally Posted by sammi jo
for music on regular CDs, being sampled at 44.1kHz, this means that at the higher end of the the frequency range, each cycle is only represented by very few data values (for example 10khz, thats only 4.41 samples per cycle).. such a small amount of data clearly cannot accurately represent the complex material at the top end of the frequency spectrum.
It is easy to think this, when looking on the face of it. The fact is though, that it's simply not true. As long as you have at least anything greater than two samples per period (e.g. 2.00000000001 samples is enough) of the highest frequency you want to capture, you can re-create exactly the signal you sampled. Basically, wherever your samples end up, as long as the signal conforms to the bandwidth limit criterion, there is one and only one signal that could have led to the samples being the values that they are. This can be a difficult concept to grasp.
Quote:
Originally Posted by sammi jo
Then there's the problem of aliasing as a result of using inadequate sample rates... see Nyquist Theorem.. To counter the (nasty) effects of aliasing, as well as smoothing out the spiky waveforms formed as a result of "not enough data", designers of digital audio equipment have included anti-aliasing digital filters into the DA conversion circuits of equipment. This is obviously far less of an issue with SuperAudioCDs since 2.8 MHz sample rates are high enough to avoid both the aliasing and "not enough data" problems and artifacts. But with ordinary CDs and (to a lesser extent) DVD audio, this is still an issue.. and the effects of digital filters on the program material, especially in cheap massproduced equipment, can be severe, especially regarding the capability of the equipment to reproduce the phase angle of the waveform, over the entire audio frequency range. Considering that we perceive directional information of a sound via the higher frequency components of sound, and these digital filters are tuned to the higher end of the frequency spectrum, then it would follow that the spatial information contained in a stereo audio recording would be an element to suffer. Just speculating here, but maybe thats why on that vinyl vs. CD demo I heard, the stereo image was so much better defined on the analog rather than the digital version.
Indeed, anti-aliasing filters can be a problem if poorly implemented.
Those who feel that CD is not adequate may be interested in this recent Audio Engineering Society Journal paper that shows that the difference between CD and higher-resolution audio formats (DVD-Audio and Super-Audio CD) is not audible. I am not aware if the paper is available for free anywhere.
Both digital and analogue recording to vinyl add noise (quantisation with dithering is exactly the same as adding random noise to the signal, that you clearly do not understand this Splinemodel and I are unable to help), vinyl adds much, much more and the digital signal is therefore more accurate.
No, it's not the same at all. With a vinyl recording, you get noise from the medium. With dithering, you're intentionally adding noise to compensate for inaccuracies/quantization error. You're still likely to hear a digital recording as being better because there is so much less unintentional noise. But the the analog recording is capturing the actual sound wave. This can make an audible difference at times...even if very small.
Quote:
Have you ever taken the output of a record deck, and passed it through a 16-bit/44.1 kHz AD-DA loop and compared the two pathways (i.e. "straight from record deck -> amplifier -> speaker" and "record deck->44.1 kHz/16 bit AD converter -> DA converter -> amplifier -> speaker)? I'm willing to bet you that in a properly controlled double-blind experiment you wouldn't be able to tell the difference with 95% confidence or better.
Of course I haven't. Have you? It's not a valid comparison anyway. You're merely testing the D/A converter. That's not the only issue with digital.
Quote:
You can't take a commercial vinyl recording and commercial CD recording of allegedly the same material and use that as the comparison. CDs go through a mastering process that sadly in many cases makes a mockery of CD's enhanced dynamic range. Some folk just think "louder=better" and compress the hell out of the original signal's dynamic range (making the whole lot louder) before pressing it to CD.
"You can't compare the gas mileage of a hybrid Toyota Camry and a non-hybrid Toytota Camry because the manufacturer limits the fuel efficiency of the hybrid for reasons X, Y and Z. Theoretically, it's way better though."
Also, I doubt you'll find many classical recordings that are compressed to all hell. I'm sure it happens a lot with pop, rock, etc.
Quote:
Sounds like the test wasn't double-blind. Sadly this means it had/has no validity whatsoever.
This isn't a fucking laboratory. Sammi and I both had similar experiences. Many others have as well.
Quote:
CD is not a severe compromise. It is perfectly adequate.
Absolutely, I agree. But it's not as good as many think.
Quote:
It is easy to think this, when looking on the face of it. The fact is though, that it's simply not true. As long as you have at least anything greater than two samples per period (e.g. 2.00000000001 samples is enough) of the highest frequency you want to capture, you can re-create exactly the signal you sampled. Basically, wherever your samples end up, as long as the signal conforms to the bandwidth limit criterion, there is one and only one signal that could have led to the samples being the values that they are. This can be a difficult concept to grasp.
Indeed, anti-aliasing filters can be a problem if poorly implemented.
Those who feel that CD is not adequate may be interested in this recent Audio Engineering Society Journal paper that shows that the difference between CD and higher-resolution audio formats (DVD-Audio and Super-Audio CD) is not audible. I am not aware if the paper is available for free anywhere.
So your position is honestly that there is no difference between S/A-CD, DVD-Audio and CD? Why, because of this?
Quote:
The authors report on a series of double-blind tests comparing the analog output of high-resolution players playing high-resolution recordings with the same signal passed through a 16-bit/44.1-kHz ?bottleneck.?.....
The test results show that the CD-quality A/D/A loop was undetectable at normal-to-loud listening levels, by any of the subjects, on any of the playback systems. The noise of the CD-quality loop was audible only at very elevated levels.
First, we're not just talking about noise. We're talking about more subjective factors...seperation, depth, warmth, tonal quality, etc.
Secondly, the noise was detectable, only at "very elevated" levels. Well what does that mean exactly? What was the source material?
Third: I don't think running the signal through a "bottleneck" is the same as listening to two different formats on the same equipment. The high resolution recording simply contains more information, particularly in terms of the sampling rate and the bit depth. I find it hard to believe that trying to "block" some of that information through a A/D/A conversion (bottleneck) would be exactly the same as listening to a source that is actually recorded differently.
No, it's not the same at all. With a vinyl recording, you get noise from the medium. With dithering, you're intentionally adding noise to compensate for inaccuracies/quantization error. You're still likely to hear a digital recording as being better because there is so much less unintentional noise. But the the analog recording is capturing the actual sound wave. This can make an audible difference at times...even if very small.
Tell me: which one of us is it that has been studying Electronics since they were 13, is now 28, has the top first in their year in a Masters Degree in Electronic Engineering and is currently doing a PhD in Audio Electronics?
This is not a matter of opinion, it is a matter of fact. I am right and you are wrong. You can say what you like, it won't change the facts. Either you can go and study this until you understand it, or you can continue to blather nonsense.
Quote:
Originally Posted by SDW2001
"You can't compare the gas mileage of a hybrid Toyota Camry and a non-hybrid Toytota Camry because the manufacturer limits the fuel efficiency of the hybrid for reasons X, Y and Z. Theoretically, it's way better though."
Way to prove you didn't understand the point I was making.
Quote:
Originally Posted by SDW2001
This isn't a fucking laboratory. Sammi and I both had similar experiences. Many others have as well.
Without performing a test double-blind, the observer-expectancy-effect/placebo heavily influences what you perceive. Someone tells you you are listening to something different makes it much more likely you will "hear" differences.
Quote:
Originally Posted by SDW2001
So your position is honestly that there is no difference between S/A-CD, DVD-Audio and CD? Why, because of this?
Yes. I've read the paper (and understand it). Have you? The test was well controlled and used a number of "experienced listeners". No one could hear the difference. It is probably worth mentioning at this point that I was very surprised by this. I was formerly of the opinion that some experienced listeners probably could hear the difference due to the aforementioned anti-aliasing filter problems and the improved dynamic range.
Quote:
Originally Posted by SDW2001
First, we're not just talking about noise. We're talking about more subjective factors...seperation, depth, warmth, tonal quality, etc.
The paper isn't just a study of the audibility of noise. The only reason noise is mentioned in the abstract is that the higher noise floor of the CD-quality A/D/A loop was audible at "very elevated" levels. The test in general was a simple comparison between the audio quality (encompassing all those things you mentioned) of high-resolution digital and CD-resolution digital. Not a single one of the listeners could hear the difference. Many tests were conducted over the course of a year, with several set-ups and many different recordings.
Quote:
Originally Posted by SDW2001
Secondly, the noise was detectable, only at "very elevated" levels. Well what does that mean exactly?
It means that when they turned the volume up real loud, listeners could tell when the A/D/A loop was switched in to the audio path during quiet passages, because they could hear the higher noise of that path. However, this was at volume levels that caused physical pain during the not-quiet passages.
Quote:
Originally Posted by SDW2001
Third: I don't think running the signal through a "bottleneck" is the same as listening to two different formats on the same equipment. The high resolution recording simply contains more information, particularly in terms of the sampling rate and the bit depth. I find it hard to believe that trying to "block" some of that information through a A/D/A conversion (bottleneck) would be exactly the same as listening to a source that is actually recorded differently.
But that's exactly what happens (or should happen) during the recording process (i.e., recording and mastering is performed at 24 bit/192 kHz, which is then converted to 16-bit/44.1 kHz, stored on CD, converted back to analogue, amplified and connected to speakers). The test shows that 16-bit/44.1 kHz is indistinguishable from 24 bit/192 kHz or SACD modulation.
The problem with CD is, as I mentioned earlier, idiots who master the recording by compressing the hell out of it. That's got nothing to do with 16-bit/44.1 kHz. It is one of the reasons that vinyl can "sound better". The actual medium though, is vastly inferior to CD in terms of signal fidelity.
Without performing a test double-blind, the observer-expectancy-effect/placebo heavily influences what you perceive. Someone tells you you are listening to something different makes it much more likely you will "hear" differences.
That experience to which I was referring was not double blind, of course. That would have been impossible, given the background noise that is always present in a vinyl recording, which gives the show away before any music starts. However, we (the audience) didn't know which medium was first up, but as soon as the music started on the vinyl version (the "B" in the A-B test), it was immediately apparent that the vinyl sounded more *real*, or *musical*. Of course the vinyl wasn't technically as perfect, given the pops, blats, clicks and surface noise from the vinyl, whereas the music from the CD emerged out of a background of digital zeros.. i.e. absolute silence (ignoring whatever noise was being generated by the amplifier, which in this case was inaudible). Yes, I know that "real","musical", "presence", "sense of space" etc etc are subjective listening terms, not scientifically quantifiable, but isn't that what listening to music is all about, and whichever equipment delivers the most "musical" reproduction is surely the most desirable? Yes, the CD has a larger dynamic range, delivers zero noise, and according to the theory (ie a perfect digital representation of the original waveform), should have beaten any vinyl competition, especially given the nature and cost of the CD playback equipment and DA converters ($ many thousands!). But that is not the impression that anybody in the audience got. I realize this is getting into dodgy ground here, and an area I am not very comfortable with, especially as a skeptic... but I know my ears fairly well, and have been to more live orchestral concerts than I can count.... but the vinyl was the clear musical winner. One can only speculate re. the cause, and that doesn't get anyone anywhere.
Regarding the claim that 44.1kHz at 16 bit resolution can deliver, in theory, audio "perfection", why has Digidesign been pushing 192kHz sample rate at 24bits for their ProTools system, and why has the audio industry even bothered developing SACD technology, when according to the stated theory, the 8 extra bits and 27,955,900 extra samples per second is "excess baggage"?
That experience to which I was referring was not double blind, of course. That would have been impossible, given the background noise that is always present in a vinyl recording, which gives the show away before any music starts.
It's not impossible to show that the CD medium itself is not a problem - you record the output of a vinyl deck, digitise it, record it to CD and do the A/B comparison that way - the noise of the vinyl is stored on the CD and then there's no noise floor difference to give the game away.
I say again that supposedly "the same" commercial recordings can be mastered completely differently for vinyl and CD - often with idiot decisions made on the CD front, resulting in better sound from vinyl. But that's not the fault of the CD medium itself.
Of course, it does boil down to if vinyl to you sounds better, it's a no-brainer that that's what you should listen to. Just don't kid yourself that it's because vinyl as a medium gives you "more" of the music, or a "more accurate" waveform.
Really, you should be getting mad at mastering folk ruining so many CD recordings - it should be giving you better sound quality but often it fails through no fault of the medium itself.
Quote:
Originally Posted by sammi jo
Yes, the CD … delivers zero noise
Not zero noise, just a few orders of magnitude lower than vinyl.
Quote:
Originally Posted by sammi jo
Regarding the claim that 44.1kHz at 16 bit resolution can deliver, in theory, audio "perfection", why has Digidesign been pushing 192kHz sample rate at 24bits for their ProTools system, and why has the audio industry even bothered developing SACD technology, when according to the stated theory, the 8 extra bits and 27,955,900 extra samples per second is "excess baggage"?
As I alluded to earlier, I don't think it was that unreasonable to expect higher-resolution mediums to deliver higher-quality audio.
In comparison to vinyl, the CD medium is superior in every way but didn't seem to quite reach the limits of human hearing in terms of dynamic range, and the anti-aliasing filters required could potentially cause issues (mentioned earlier in this thread) if not implemented properly.
Increasing bit-depth delivers higher dynamic range and increasing the sampling rate makes it possible to implement analogue anti-aliasing filters with linear phase in the entirety of audible passband. Surprisingly, given the length of time higher-res formats have been around, that paper reports on the first double-blind comparison that I'm aware of between high-res and 16-bit/44.1 kHz. As it turns out, the higher-res is a waste of time.
It's not impossible to show that the CD medium itself is not a problem - you record the output of a vinyl deck, digitise it, record it to CD and do the A/B comparison that way - the noise of the vinyl is stored on the CD and then there's no noise floor difference to give the game away.
Yes, that is correct. But given the mature of the comparison, there was no time to get into all that rigmarole.
Quote:
I say again that supposedly "the same" commercial recordings can be mastered completely differently for vinyl and CD - often with idiot decisions made on the CD front, resulting in better sound from vinyl. But that's not the fault of the CD medium itself.
Agreed in principle.
So much depends on the nature of the music that is being mastered. Most modern non-orchestral material has by nature a tiny dynamic range, and is further compressed at source, in the multi-tracking, mixing and final mastering environments. If you analyze the dynamic range of (most) music that is aimed at radio play and there's barely 2 dB difference between the "quieter" and "louder" passages, at the most. When broadcast on radio, further compression is added in the controlroom... listen to some of that rare species of rock on radio that has some dynamic range, and you can hear the compression kicking in during the quieter passages... sometimes to the extent that noise encoded in older recordings, (or even the noise generated by the playback equipment in the controlroom!) is pumped up by the compressors as they attempt to keep the audio at a "flatline" level. Since most music is listened to casually as part of the furniture (as opposed to active listening), and often in unfriendly listening environments such as in vehicles, and other places with a high ambient noise level (or other distractions), quieter passages will be lost, or even inaudible.
Orchestral music however has a dynamic range (from 'ppp' to 'fff') of up to 75dB, and the relative difference in SPL is an essential component of the music, an essential compositional device. Different mastering techniques are employed for "classical" music. But for classical music radio broadcast it is a necessary compromise to employ some degree of compression otherwise the quieter passages will be lost to most listeners.. of course, classical recordings cover so much different varieties of music with different amounts of dynamic range, and some will suffer far more that others. However, for those who actively listen to classical music on quality audio playback systems at home, the ability to reproduce the full dynamic range is essential for the listening experience.
And yes, CD, re. ability to reproduce dynamic range, is superior to vinyl or tape.
Quote:
Of course, it does boil down to if vinyl to you sounds better, it's a no-brainer that that's what you should listen to. Just don't kid yourself that it's because vinyl as a medium gives you "more" of the music, or a "more accurate" waveform.
Oh agreed. But I am still curious as to the "why" of vinyl, or even better, half-inch analog tape at 30ips. Its a bit like state of the art tube amplifiers "holding their ground" compared to state of the art transistor amplifiers, yet the quoted technical specifications of such tube amps are (as a rule) vastly inferior to their transistorized counterparts....
Quote:
Really, you should be getting mad at mastering folk ruining so many CD recordings - it should be giving you better sound quality but often it fails through no fault of the medium itself.
There are plenty of cowboys out there (really), in amongst a small minority of highly skilled mastering engineers. And there are many recordings out there which sound "brittle and edgy" (almost unlistenable for me), especially amongst recordings that used early digital equipment, from the late 1970s and early 1980s.
Quote:
As I alluded to earlier, I don't think it was that unreasonable to expect higher-resolution mediums to deliver higher-quality audio.
In comparison to vinyl, the CD medium is superior in every way but didn't seem to quite reach the limits of human hearing in terms of dynamic range, and the anti-aliasing filters required could potentially cause issues (mentioned earlier in this thread) if not implemented properly.
Increasing bit-depth delivers higher dynamic range and increasing the sampling rate makes it possible to implement analogue anti-aliasing filters with linear phase in the entirety of audible passband. Surprisingly, given the length of time higher-res formats have been around, that paper reports on the first double-blind comparison that I'm aware of between high-res and 16-bit/44.1 kHz. As it turns out, the higher-res is a waste of time.
I understand what you are saying, but are you now implying that 44.1/16[I] is a compromise, re. the need for a "perfect" brickwall AA filter to allow for an exact digital copy of the waveform at the 44.1kHz sample-rate (which is the absolute minimum needed to reproduce the full audio spectrum) and bit depth?
Regarding the claim that 44.1kHz at 16 bit resolution can deliver, in theory, audio "perfection", why has Digidesign been pushing 192kHz sample rate at 24bits for their ProTools system, and why has the audio industry even bothered developing SACD technology, when according to the stated theory, the 8 extra bits and 27,955,900 extra samples per second is "excess baggage"?
I've been out of the audio recording loop for nearly a decade but here's my 2 cents.
1) It used to be conventional thinking that when you start mixing tens to hundreds of 16 bit/44.1 tracks together you need greater bit precision - not ultimately for the dynamic range of the music, but because of the mathematics involved in adding what is essentially a collection of 1's and 0. For a long time 24 bit devices, while theoretically better, had worse recording/reproduction characteristics than 16/44.1 - maybe thats why the ultimate pro gear didn't move to 24 bits right away.
It is also beneficial to have headroom when recording because digital clipping (excess signal over the available bandwidth) is quite nasty compared to analog clipping - which is quite nice - and thats part of the reason vinyl has a distinctive character - its analog clipping - which is a form of compression in itself as it is non-linear and harmonic, while digital clipping is absolute, harsh and disharmonic. Headroom is very important, because naturally played instruments aren't compressed by default when recording digitally, so anything over 0db will cause nasty distortion. When recording analog, signals over 0db are non-linearly attenuated because of the magnetic saturation of the medium, which is like I say just a 'nice' kind of distortion.
Even then you still take your master mix of several tens of 24/192 tracks and dither the final output to 16 bits. This isn't done because it sounds better per se, but because of the maths i mentioned. All you are trying to do is create a 2 channel mix, so you need to do the maths at 24/192 because you have lots of individual tracks - but the final 2 channel is as perfectly good at 16/44.1 or 48. This doesn't really change the waveform of the finished music*, but just scales it to fit in a 16 bit resolution. I dont know if it is still the case, but early systems used to just drop the few least significant bits to scale to 16 bit - because you simply cannot hear them - i suspect these days its a bit more sophisticated.
*i know it does theoretically, but due to nyquists theorem and a half decent filter you cannot tell the difference. Back in the day I remember reading in the journal Sound on Sound - I had a pile of them from floor to ceiling - hehe! that research was ongoing into the psychoacoustic effect of frequencies above the threshold of human hearing, ie 20khz being percievable - not as sound, but in the effects it has on sound in the range of hearing. I dont know what the outcome of this was - and I remember Pioneer releasing cd players that interpolated higher than human range frequencies - up to about 60khz IIRC, but i dont think that ever caught on.
Microphones do not have the range that 24bits can store - and if you look at the most expensive mics in the world - it really isn't the quality of reproduction of the source that makes them sought after - but the analog clipping and shaping they perform. Tubes in particular sound lovely - but thats because they are non-linear compressors, saturators and distorters in effect.
It used to be the case, Im not sure if it still is, that even though the ADC's were rated at 24 bits, the actual electronics involved could not actually perform at 24 bits anyway. The SNR ratio is so high - IIRC something like 132db that you're actually hitting the limits of the theoretical maximum attainable SNR which is ultimately limited by the random noise of molecular motion as governed by quantum physics.
2) Every computerised system i've ever come across suffers from what I would call the marketing BS factor. Ignorant people like big numbers. Thats why they rate monitors at 5000:1 contrast, 700 cd/m3 - even though you'd have no retinas left if you used it like that. Its why DDR3 ram is rated at 2ghz though it performs barely 5% better than 667 mhz DDR2. Its why people think 16x PCIe performs twice as good as 8xPCIe for graphics when the difference is less than 1%
'The industry' largely develops things because its something new to sell you at a greater margin. Now that CD players can be had for $10 and CD sales are declining, pushing a new format and having the public re-invest their catalog in the new format is a money spinner. 24/192 is not 4x better than CD, maybe it is 5% better for the 1% of people who can tell the difference, but cue the law of diminishing returns...its not because of the quality at all.
Its all just shit. Mr H is spot on. There is a quality about vinyl that makes it attractive, but thats because of what it 'adds' to the recorded wave in terms of analog clipping, saturation and distortion, not because it reproduces anywhere as good as CD.
If you want to hear something truly amazing, get hold of an SPL vitalizer and run your CD's and vinyl through that. It gets tiring on the ears and was much overused - but just as an exercise its good fun.
Another thing - vinyl is absolutely rubbish for off centre deep bass sounds - all kinds of phase distortions from the stylus not being able to accurately follow the grooves in the disc
But the the analog recording is capturing the actual sound wave. This can make an audible difference at times...even if very small.
Im afraid that is a popular misconception. While we all know the limitations of digital recording, to say that analog captures the actual sound wave is completely false.
Analog circutry and recording has all of the things built in by default that you deride against digital recordings. Compression, saturation, distortion. its unavoidable by the laws of physics. Much research has gone into reproducing the characteristics of analog limitations and distortions so that we can add it to digital recordings to make it sound more pleasing. If in some strange parallel universe digital had come before analog, then analog would be totally slated as an accurate recording medium,
It may sound more pleasing - but thats 180 degrees to the complaint. Digital has its own characteristics of compression, saturation and distortion, but the reason you like analog is the way it distorts!
The only way to hear what the musician intended is to hear it live - but thats not really the case because the venue will add its own form of wave-shaping to the sound of the instruments.
Thats what I was referring to re. CDs being a severe compromise to start with (that is before they are reduced yet further by one of compression algorithms to Mpeg3 formats!). Considering that the human audio frequency range is (approx.) 20hz - 20khz -- for music on regular CDs, being sampled at 44.1kHz, this means that at the higher end of the the frequency range, each cycle is only represented by very few data values (for example 10khz, thats only 4.41 samples per cycle).. such a small amount of data clearly cannot accurately represent the complex material at the top end of the frequency spectrum. design... as if that section of the device is an afterthought... ask any serious audio designer. Unless one is paying top dollar, one can guarantee that the analog audio components and circuit design is as cheap as the manufacturer can get away with. Even some "top end" digital studio consoles use opamps and associated circuitry with could never be honestly described as truly "high fidelity". So, when it comes to the "weakest link in the chain" which determines the audio quality for the listener, there are so many places where the signal is compromised. "SuperAudioCDs" will only reproduce what is there in the first place, and if the program material is sonically compromised, then that compromise (error) will be more accurately reproduced as well!
'Scuse the length of this reply!
I dont think its valid to say...20 years ago I compared this to that.
20 years ago digital recording was in its infancy, while vinyl was at a near prime. The spec for CD's hasn't changed, but the technology for getting those bits onto the disc was rather crude.
Its a bit like comparing DVD to Blu-ray. Yes the first gen of blu-ray might take 30 seconds to display the root menu, and skip a few frames, sluggish fast-forward etc, but then the computational power to do this is several magnitudes higher than DVD. Give it a few years for Blu-Ray players to attain the power and all the faults of the new medium go away.
But the potential for blu-ray is and always was set right from the beginning.
Also youre complaining that the cheap end of digital devices have crap filters etc - well the cheap end of analog is particularly crap if I recall - far worse than the crap end of digital reproduction.
The other thing when critisizing distortion -remember that distortion creates harmonics. By default a harmonic is a multiple of the fundamental. So even if you're distorting a 20khz wave - something I doubt you can hear anyway - the first harmonic of this is going to be 40khz - something you definately wont hear. While I agree that the top end of audio is compromised by recording digitally, when done properly, it isn't something to worry over. Also high frequencies (and low frequencies) aren't 'heard' in the same way as mid-ranges. Its quite easy to determine the timbre of a sound in the mid ranges, but when it comes to high-frequencies (and low), what is detected is more of a 'presence of energy' than a timbre. Therefore even mildly distorted high frequency waves are not detectable by the ear to be distorted - and the brain has alot to do with this too - because you are only sensing presence of energy, rather than timbre. IIRC the switch from timbre to 'presence' occurs quite low, at about 5khz, and bear in mind that any frequency you hear above say 5khz is likely to be a high-order harmonic created from fundamentals around 100hz to 3khz on average, the amount of timbral distortion you will perceive above 10khz is going to be near negligable.
Tell me: which one of us is it that has been studying Electronics since they were 13, is now 28, has the top first in their year in a Masters Degree in Electronic Engineering and is currently doing a PhD in Audio Electronics?
This is not a matter of opinion, it is a matter of fact. I am right and you are wrong. You can say what you like, it won't change the facts. Either you can go and study this until you understand it, or you can continue to blather nonsense.
Way to prove you didn't understand the point I was making.
Way to be an ass. If you can't discuss the topic at hand without being condescending and rude, then go away.
As for not understanding your point, you're wrong. I think it's you who doesn't understand the simple point that a CDs theoretical performance does not equate to a CDs typical performance. Are there idiots mastering CDs? Absolutely. According to you, they may all be idiots. The fact is though...it doesn't matter. We can buy what we can buy.
Quote:
Without performing a test double-blind, the observer-expectancy-effect/placebo heavily influences what you perceive. Someone tells you you are listening to something different makes it much more likely you will "hear" differences.
I didn't know what I was listening to or that it was different. In fact, I had assumed it was a CD.
Quote:
Yes. I've read the paper (and understand it). Have you? The test was well controlled and used a number of "experienced listeners". No one could hear the difference. It is probably worth mentioning at this point that I was very surprised by this. I was formerly of the opinion that some experienced listeners probably could hear the difference due to the aforementioned anti-aliasing filter problems and the improved dynamic range.
So you're basing your position on a single study. Awesome.
Quote:
The paper isn't just a study of the audibility of noise. The only reason noise is mentioned in the abstract is that the higher noise floor of the CD-quality A/D/A loop was audible at "very elevated" levels. The test in general was a simple comparison between the audio quality (encompassing all those things you mentioned) of high-resolution digital and CD-resolution digital. Not a single one of the listeners could hear the difference. Many tests were conducted over the course of a year, with several set-ups and many different recordings.
Yes, but was not a test of actual recordings played on typically used equipment. It was a test of the loop's effect, and nothing more. A more effective and realistic test would be to compare two recordings in different formats, because that's what people are able to purchase anyway. I'm sure some recordings would show more of a difference than others because of the differences in mastering. Even if the differences heard are in the mastering, those differences are still there.
Quote:
It means that when they turned the volume up real loud, listeners could tell when the A/D/A loop was switched in to the audio path during quiet passages, because they could hear the higher noise of that path. However, this was at volume levels that caused physical pain during the not-quiet passages.
But the difference was there.
Quote:
But that's exactly what happens (or should happen) during the recording process (i.e., recording and mastering is performed at 24 bit/192 kHz, which is then converted to 16-bit/44.1 kHz, stored on CD, converted back to analogue, amplified and connected to speakers). The test shows that 16-bit/44.1 kHz is indistinguishable from 24 bit/192 kHz or SACD modulation.
Is all or the majority of recording actually done at 24 bit/192KHZ? Also, what about recording at much higher sampling rates, such as 2.8MHZ?
Quote:
The problem with CD is, as I mentioned earlier, idiots who master the recording by compressing the hell out of it. That's got nothing to do with 16-bit/44.1 kHz. It is one of the reasons that vinyl can "sound better". The actual medium though, is vastly inferior to CD in terms of signal fidelity.
Even if so, it is what it is. And once again, I don't know that vinyl "sounds better" overall. I simply know that there are many, many people who've had the same experiences sammi and I have. They listened to something and thought "wow, that's got such warmth and depth and space to it....what a great CD." Then they found out it was vinyl.
Also high frequencies (and low frequencies) aren't 'heard' in the same way as mid-ranges. Its quite easy to determine the timbre of a sound in the mid ranges, but when it comes to high-frequencies (and low), what is detected is more of a 'presence of energy' than a timbre. Therefore even mildly distorted high frequency waves are not detectable by the ear to be distorted - and the brain has alot to do with this too - because you are only sensing presence of energy, rather than timbre. IIRC the switch from timbre to 'presence' occurs quite low, at about 5khz, and bear in mind that any frequency you hear above say 5khz is likely to be a high-order harmonic created from fundamentals around 100hz to 3khz on average, the amount of timbral distortion you will perceive above 10khz is going to be near negligable.
But those "energies" are perceived, and may well result in our perceptions of space, warmth, depth, etc.
Im afraid that is a popular misconception. While we all know the limitations of digital recording, to say that analog captures the actual sound wave is completely false.
Uh, OK.
Quote:
Analog circutry and recording has all of the things built in by default that you deride against digital recordings.
I'm not deriding anything.
Quote:
Compression, saturation, distortion. its unavoidable by the laws of physics.
I agree.
Quote:
Much research has gone into reproducing the characteristics of analog limitations and distortions so that we can add it to digital recordings to make it sound more pleasing.
If in some strange parallel universe digital had come before analog, then analog would be totally slated as an accurate recording medium,
Yes, but we are adding those things, which simply cannot be the same as them existing naturally. In fact, I believe it was you who stated that wrt clipping in an earlier post.
Quote:
It may sound more pleasing - but thats 180 degrees to the complaint. Digital has its own characteristics of compression, saturation and distortion, but the reason you like analog is the way it distorts!
First, I don't prefer analog overall (for about the 7th time now, actually). Secondly, even if one prefers analog because of the natural distortion, it's still a difference that's worth noting. It's not the same as digital.
Quote:
The only way to hear what the musician intended is to hear it live - but thats not really the case because the venue will add its own form of wave-shaping to the sound of the instruments.
well SDW, I was addressing things in the sense of the general complaint about analog vs digital - if you're fine with it good on you!
Lets address another thing - the presence of energy that you perceive rather than hear. First off I dont fully know the answer to this question, however I think it is possible to look at other systems that I do know about to get a good guesstimate about how likely it is that ultrasonic waveforms have a perceptible effect on normal sounds.
A quick look on Wiki confirms that ultrasonic noises, those above 20khz can have an 'impression' on the mind - if the sound is 'wired' directly to the skull and bypasses the bones in the ear.
The question is - is it this effect that is responsible for the pleasing characterists of vinyl vs digital? I think the answer is no.
Lets go back a step and consider the sampling of every individual note of a piano at 24/196 for a massive multisampled instrument to get a clear picture of the variables at play.
The problem with sampling every note of a piano is one of resonances between the strings. In effect all other things being equal, striking one note causes other strings to vibrate in sympathy, but only certain other ones. The way this is worked out is that if a string has an overtone in common of one of the harmonics, a resonant wave will be set up.
So if you pluck a string say 200hz, then strings tuned to 400hz, 600hz, 800hz etc will resonate. However, you have to look at the waveform of the fundamental to see how much energy is transfered to the other strings. Assuming a sawtooth wave - which has all odd and even harmonics falling away exponentially in energy from the fundamental, you can visualize that it would take one hell of a 200hz wave to get any meaningful resonances past the 10th harmonic.
That being said, after you have sampled every note of the piano, (which incidently does not look anything like a sawtooth wave, it has far fewer harmonics, you might think you have the perfect sounding piano - all resonances included) . Nope!
The voodoo thing ... is when you strike 2 notes together on a real piano - it sets up a different set of resonant interplay between the strings than you would verifyably get if you hit both notes seperately. And offcourse you have not sampled these variations...You cant accurately sample a piano - unless you take a sample of every single combination of the 88 keys - and that is not possible.
However - we have looked at this in reverse of what we are trying to find out. We have been hitting a low note and finding out if higher tuned strings resonate. They do. Now what happens if you hit a high note? do lower tuned strings resonate?
They do....But to what extent?
But its more complex. In order for a lower note to resonate in sympathy with a higher note, it has to have a harmonic sequence that at some point coincides with the fundamental or one of the harmonics of the high note.
So for instance, it I strike a virtual 1000hz sawtooth, my first harmonic is going to be 2000hz - but at half the energy. Theoretically, a 500hz virtual string will have a 3rd harmonic at 2000hz so they can resonate, however the 3rd harmonic of a sawtooth is 1/3 the energy of the fundamental, so even in the best theoretical example you can imagine, the transfer of resonant energy is pretty miniscule. Especially when you consider that sound is measured on a logarathmic scale BUT heard only as one continuous wave by the ear - which is deconstructed in the brain - which is really voodoo.
But lets take this too our digital ultrasonic example. Can say a 40khz wave cause resonances in audible sounds that to any practical extent can be perceived by a human - either consciously or subconsciously.
Starting with the basics. Any wave that is created at 40khz is going to be by default a very high order harmonic of something. That means it contains very little energy. Sawtooths are extreme examples - most instruments produce nowhere near as much harmonics as a sawtooth - or square wave.
Any ultrasonic harmonic produced by a natural instrument is going to be extremely low energy. Its potential for causing a resonance in the audible spectrum is therefore extremely low. Then factor in that an audible sound has to have an harmonic in common with the ultrasonic sound to resonate, and again by default, this harmonic is going to be of a very high order with a miniscule amount of energy attached to it. To all extents and purposes the resulting cross product resonance is basically nothing.
Now if you take a high frequency pure-tone audible wave in complete isolation and resonate it with a single tone ultrasound and measure the distortion very accurately - in labratory settings - you will find a very small detectable resonance - and this is what I believe the hi-fi freaks have latched on to. They're shouting "look its there - ITS THERE!!!' -Im missing something yadda yadda yadda!!!"
But its not there, it was created by a special electronic circuit designed to output things you cannot hear - and more importantly - really would not want to hear.
What they're missing apart from a few marbles, is that this is not the realworld. There is no music id want to listen too that contained nothing but 20khz puretones modulated with ultrasonic frequencies - and that the whole test is designed to show what happens in the very worst case scenario - again something that never happens in the realworld.
Then factor in that even in the very best recording studios, by the time the ultrasonic wave has passed through all the equipment that is needed to record and duplicate the sound in the first place - that is IF the microphone even picked it up in the first place.....It simply does not exist in the vinyl recording anyway!
It is not the Low Pass Filters, Bit rate and sampling frequency (ie lack of ultrasound) that are the issues in recordings made to CD's - just as it is not the presence of ultrasonic sound that makes vinyl sound the way it does.
The pleasurable warmth of vinyl recording is because of magnetic saturation, non-linear distortion, magnetic hysterisis, eddy currents, frequency responce of capacitors and inductors in the circuit....and a million other things - ALL OF WHICH SERVE TO DEVIATE THE RECORDED WAVEFORM FROM THE ORIGINAL- yes it sounds nice, but thats the black art form of the audio designer.
From a technical, accuracy, perspective, when done right, CD's reproduce the original waveform much more accurately, subject to thier characteristic distortions.
The funniest thing is, what it comes out of, is and will always be an analog device. The Loudspeaker. When you physically reproduce the sound on your loudspeaker, the cone has to move between the positions of any two samples anyway, thus smoothing the sampled waveform of its sampling frequency, bit depth and step distortion- thus interpolating all the missing *theoretical* perceived energy you cant hear anyway - anyway!!!
Yes, but we are adding those things, which simply cannot be the same as them existing naturally.
i'll address this point in particular.
What they're doing is analysing the reproductive characteristics of analog devices using probably not much more than simple sine waves, measuring how the analog circuitry distorts the wave over the entire audible spectrum, creating a computer model of frequency response and adding back in the same kinds of distortion.
So which manufacturers model of analog circuitry do you analyse and model? What turntable maker sounds the absolute best??? very subjective....
It will never sound exactly the same as vinyl. However there is a point that will be reached where the decibel level of missing distortion or error is so small as to be indistinguishable.
However - consider this thought experiment.
If vinyl sounds so good, why would the clued-up audio engineers simply not record their best original source to vinyl and sample the vinyl for release on CD?
What would happen if you did it vice-versa.
Which would produce the most accurate reproduction of the original sound as the musician intended?
So anyway, enough of this thread. The conclusion....
First a dislaimer. I have not studied this for a long time, I cant promise i've outlined every principle 100% accurately.
Secondly, I dont really care either way in the analog vs digital debate, a good system either analog or digital can sound fabulous.
Thirdly, if you prefer either principle - why cant it just be 'because I/you do'. From what I can fathom, most audiophiles completely invent the most ridiculous psuedosciencejunk for their reasoning.
Fourth - lets not get started on 99.9% oxygen free cabling...
well SDW, I was addressing things in the sense of the general complaint about analog vs digital - if you're fine with it good on you!
I am fine with it. As I've said, digital wins hands down for me. The advantages are enormous. All I'm saying is that many people analog recordings have a certain warmth, depth, space (etc.) that digital ones don't have...and I tend to agree with them, at times.
Quote:
Lets address another thing - the presence of energy that you perceive rather than hear. First off I dont fully know the answer to this question, however I think it is possible to look at other systems that I do know about to get a good guesstimate about how likely it is that ultrasonic waveforms have a perceptible effect on normal sounds.
A quick look on Wiki confirms that ultrasonic noises, those above 20khz can have an 'impression' on the mind - if the sound is 'wired' directly to the skull and bypasses the bones in the ear.
The question is - is it this effect that is responsible for the pleasing characterists of vinyl vs digital? I think the answer is no.
I don't know...I was just throwing it out there. It might, it might not. I just think there has to be an explanation. I find it unlikely that all the people out there who hear analog as being different/better are victims of a placebo or nostalgia effect.
Quote:
Lets go back a step and consider the sampling of every individual note of a piano at 24/196 for a massive multisampled instrument to get a clear picture of the variables at play.
The problem with sampling every note of a piano is one of resonances between the strings. In effect all other things being equal, striking one note causes other strings to vibrate in sympathy, but only certain other ones. The way this is worked out is that if a string has an overtone in common of one of the harmonics, a resonant wave will be set up.
That's much more complex than you've made it out to be. First, the piano is dampened. Secondly, the overtones in the string are going to be present regardless of any other sympathetic vibrations.
Quote:
So if you pluck a string say 200hz, then strings tuned to 400hz, 600hz, 800hz etc will resonate. However, you have to look at the waveform of the fundamental to see how much energy is transfered to the other strings. Assuming a sawtooth wave - which has all odd and even harmonics falling away exponentially in energy from the fundamental, you can visualize that it would take one hell of a 200hz wave to get any meaningful resonances past the 10th harmonic.
OK....
Quote:
That being said, after you have sampled every note of the piano, (which incidently does not look anything like a sawtooth wave, it has far fewer harmonics, you might think you have the perfect sounding piano - all resonances included) . Nope!
Actually, that's a poor example because pianos are tuned under Equal Tone Temperament (mathematically imperfect, but equal intervals). So sampling every note is fine, but the notes are not mathematically in tune.
Quote:
The voodoo thing ... is when you strike 2 notes together on a real piano - it sets up a different set of resonant interplay between the strings than you would verifyably get if you hit both notes seperately.
I'm not sure that's true. Where did you get that from?
Quote:
And offcourse you have not sampled these variations...You cant accurately sample a piano - unless you take a sample of every single combination of the 88 keys - and that is not possible.
There is no need for that because every interval is equal but imperfect. All one needs to do is sample each of the 88 pitches. The combinations don't matter. What you say would be true (I think) if the piano was tuned with Just Intonation or another system.
Quote:
However - we have looked at this in reverse of what we are trying to find out. We have been hitting a low note and finding out if higher tuned strings resonate. They do. Now what happens if you hit a high note? do lower tuned strings resonate?
They do....But to what extent?
But its more complex. In order for a lower note to resonate in sympathy with a higher note, it has to have a harmonic sequence that at some point coincides with the fundamental or one of the harmonics of the high note.
So for instance, it I strike a virtual 1000hz sawtooth, my first harmonic is going to be 2000hz - but at half the energy. Theoretically, a 500hz virtual string will have a 3rd harmonic at 2000hz so they can resonate, however the 3rd harmonic of a sawtooth is 1/3 the energy of the fundamental, so even in the best theoretical example you can imagine, the transfer of resonant energy is pretty miniscule. Especially when you consider that sound is measured on a logarathmic scale BUT heard only as one continuous wave by the ear - which is deconstructed in the brain - which is really voodoo.
But lets take this too our digital ultrasonic example. Can say a 40khz wave cause resonances in audible sounds that to any practical extent can be perceived by a human - either consciously or subconsciously.
Starting with the basics. Any wave that is created at 40khz is going to be by default a very high order harmonic of something. That means it contains very little energy. Sawtooths are extreme examples - most instruments produce nowhere near as much harmonics as a sawtooth - or square wave.
Any ultrasonic harmonic produced by a natural instrument is going to be extremely low energy. Its potential for causing a resonance in the audible spectrum is therefore extremely low. Then factor in that an audible sound has to have an harmonic in common with the ultrasonic sound to resonate, and again by default, this harmonic is going to be of a very high order with a miniscule amount of energy attached to it. To all extents and purposes the resulting cross product resonance is basically nothing.
Now if you take a high frequency pure-tone audible wave in complete isolation and resonate it with a single tone ultrasound and measure the distortion very accurately - in labratory settings - you will find a very small detectable resonance - and this is what I believe the hi-fi freaks have latched on to. They're shouting "look its there - ITS THERE!!!' -Im missing something yadda yadda yadda!!!"
But its not there, it was created by a special electronic circuit designed to output things you cannot hear - and more importantly - really would not want to hear.
What they're missing apart from a few marbles, is that this is not the realworld. There is no music id want to listen too that contained nothing but 20khz puretones modulated with ultrasonic frequencies - and that the whole test is designed to show what happens in the very worst case scenario - again something that never happens in the realworld.
Then factor in that even in the very best recording studios, by the time the ultrasonic wave has passed through all the equipment that is needed to record and duplicate the sound in the first place - that is IF the microphone even picked it up in the first place.....It simply does not exist in the vinyl recording anyway!
I don't know. It seems to me that the analog recording (or super high quality digital recording) would be better able to include those "inaudible" tones. They may or may not have an effect on what is "heard" or even felt/perceived. It seems like one possible explanation.
Quote:
It is not the Low Pass Filters, Bit rate and sampling frequency (ie lack of ultrasound) that are the issues in recordings made to CD's - just as it is not the presence of ultrasonic sound that makes vinyl sound the way it does.
I don't entirely agree, unless you're saying that while say, an 8 bit recording is not good, a 16 bit recording will always be capable of performing at a level beyond our ability to hear.
Quote:
The pleasurable warmth of vinyl recording is because of magnetic saturation, non-linear distortion, magnetic hysterisis, eddy currents, frequency responce of capacitors and inductors in the circuit....and a million other things - ALL OF WHICH SERVE TO DEVIATE THE RECORDED WAVEFORM FROM THE ORIGINAL- yes it sounds nice, but thats the black art form of the audio designer.
All contributing factors to be sure, but not the only ones.
Quote:
From a technical, accuracy, perspective, when done right, CD's reproduce the original waveform much more accurately, subject to thier characteristic distortions.
That's where I disagree. It's simply not true. The wave is only sampled and reconstructed. Not much is lost or estimated incorrectly, but something is. The problem in the analog comes from noise and distortions in the analog recording, which are always present. The only question is in terms of degree. High quality digital (CD and better) is so good that the vast majority of people will judge it as being "more accurate" and certainly "better" than analog. I'm certainly in that camp.
Quote:
The funniest thing is, what it comes out of, is and will always be an analog device. The Loudspeaker. When you physically reproduce the sound on your loudspeaker, the cone has to move between the positions of any two samples anyway, thus smoothing the sampled waveform of its sampling frequency, bit depth and step distortion- thus interpolating all the missing *theoretical* perceived energy you cant hear anyway - anyway!!!
I know! Of course, I don't know that it totally negates the "theoretical" energy, but whatever. The point is that after all this, we're still pushing sound through a big vibrating cone.
That's also what's amazing about people that buy electronics (amazingly stupid, that is). They'll go buy a $400 receiver and a $150 pair of speakers and think they've bought something good. In reality they've gotten one step away from the worst of the worst. In my case, I'm looking at spending $1500 for a receiver and another $1000 for speakers...and I know I'll still only have a "mid-fi" product at best. Really "good" speakers start at about $5000 a pair.
What they're doing is analysing the reproductive characteristics of analog devices using probably not much more than simple sine waves, measuring how the analog circuitry distorts the wave over the entire audible spectrum, creating a computer model of frequency response and adding back in the same kinds of distortion.
So which manufacturers model of analog circuitry do you analyse and model? What turntable maker sounds the absolute best??? very subjective....
It will never sound exactly the same as vinyl. However there is a point that will be reached where the decibel level of missing distortion or error is so small as to be indistinguishable.
However - consider this thought experiment.
If vinyl sounds so good, why would the clued-up audio engineers simply not record their best original source to vinyl and sample the vinyl for release on CD?
What would happen if you did it vice-versa.
Which would produce the most accurate reproduction of the original sound as the musician intended?
Of the two...CD or whatever digital format one is using is going to be 'better." There's no question. I was just trying to point out a possible explanation for the differences heard in analog v. digital.
I don't know...I was just throwing it out there. It might, it might not. I just think there has to be an explanation. I find it unlikely that all the people out there who hear analog as being different/better are victims of a placebo or nostalgia effect.
Its very real, analog sounds different. Whether its 'better' is subjective dont you agree? Im perfectly fine with people saying they prefer the sound of analog, because what they prefer is up to them. However their reasoning is rubblish.
They prefer the sound of analog because of the way the circuitry mangles the sound!!!!
Quote:
That's much more complex than you've made it out to be. First, the piano is dampened. Secondly, the overtones in the string are going to be present regardless of any other sympathetic vibrations.
well of course its much more complex than I said, but im not going to write 20 pages....However the fundamental principles are sound. (sic) hehe. Every instrument bar a sinewave generator has harmonics and overtones, thats what gives a sound its timbre, but you already know this being a muzak teacher....
Quote:
Actually, that's a poor example because pianos are tuned under Equal Tone Temperament (mathematically imperfect, but equal intervals). So sampling every note is fine, but the notes are not mathematically in tune.
I'm not sure that's true. Where did you get that from?
There is no need for that because every interval is equal but imperfect. All one needs to do is sample each of the 88 pitches. The combinations don't matter. What you say would be true (I think) if the piano was tuned with Just Intonation or another system.
well, getting technical, its because a piano is not tuned mathmatically perfectly that makes it a pain to sample, the resonances are close, but the effect of imperfect intervals is that it creates a chorus effect, ie periodic modulated slight detuning.
ok, im just trying to explain things simply....vibration of strings, resonances, how they work, forgetting the specific mechanics of a piano, this is how it works - and as far as were concerned - trying to establish if ultrasound effects audio, then the piano string model is easy to visualize....
However - I am perfectly right when I say that you cannot accurately capture the subtle resonances of a piano when sampling it a note at a time. Here's why
If I strike a single note - I set the string vibrating, and on its own it contains its characteristic set of harmonics that define it as a piano - However part of the characteristic of a piano is the resonance that occurs when a string is struck that sets other strings vibrating, regardless of whether they are mathmatically in tune....harmonics tend to wander from mathmatical perfection as you hit higher order ones anyway.
Now say I sample a single key - lets "symbolically" call this A, and suppose strings B, C, D, E resonate with it. I record this
A---BCDE
Now I sample another single key lets call this A1---and strings B1,C1,D1 and E1 resonate with it.
Now in my sampler, I play A and A1 together - and as i've captured the resonance of these strings, Im also reproducing resonances B,B1,C,C1,D,D1 and E and E1 as you would expect.
A perfect capture of the essence of a piano??? Not quite.....
In a real piano, the soundwaves of A and A1 would modulate eachother - something like the effect of beating on closely matched tones, and another frequency and harmonic set is created, A2---B2,C2,D2,E2 which in turn causes other resonances in strings, and thus we have in effect a characteristic of the piano that has not been captured by our sampler when we sampled every key...because playing back the 2 samples simultaneously does not recreate A2,B2,C2,D2,E2.
Lets forget the effect on timbre and harmonics by hitting the keys harder or softer...
Now, not much music is simply 2 keys of a piano, so imagine the very complex interplay between numerous notes being played and the resonances created in the physical world..its just not possible to sample it one key at a time.
However, like the ultrasound effect we are trying to solve, the effect of this is so small and virtually indistinguishable, that you can sample every key of a piano and fool 99.9% of the listeners...if you are very clever and know how to process the sound to get around the pitfalls of sampling a piano. Its really not as easy as sampling every key.
Quote:
I don't know. It seems to me that the analog recording (or super high quality digital recording) would be better able to include those "inaudible" tones. They may or may not have an effect on what is "heard" or even felt/perceived. It seems like one possible explanation.
well it cant hurt to have them present can it?
Quote:
That's where I disagree. It's simply not true. The wave is only sampled and reconstructed. Not much is lost or estimated incorrectly, but something is. The problem in the analog comes from noise and distortions in the analog recording, which are always present. The only question is in terms of degree. High quality digital (CD and better) is so good that the vast majority of people will judge it as being "more accurate" and certainly "better" than analog. I'm certainly in that camp.
The 'something' that is being lost is actually - what is being added to the analog recording that makes it perceptibly sound better.
The real problem with digital is that it suffers from being too accurate and the slight distortions it does introduce are not pleasant sounding to the ear. In contrast analog is not very faithful to the source, but includes by default distortions that are pleasant to the ear.
Really, if you dont believe me that distortions make things sound better, get an SPL vitalizer or suchlike, its a psychoacoustic processor, It distorts the crap out of sounds and they take on all the properties that people are claiming make analog sound better. Better stereo seperation, high frequency distortion to add in near ultrasonic harmonics...etc. It really isn't accurate to source in any sense of the word, but it works.
Comments
I'm not trying to be adversarial. I am, however, trying to clear up a matter that is often misunderstood. Your posts demonstrate a misunderstanding of the math at work. A reproduction of a signal can be broken down into a simple equation: reproduction = original signal + noise. Noise is a loose term. It just means "unwanted signal." Vinyl recording equipment is noise prone. It is subject to EM noise, temperature, material variance, gravity variance, and just about everything else. Digital recording isn't.
It's very difficult to reference audio reproductions to the original source, since ultimately some noise-prone delay feature is required to compare the source to the reproduction. However, it's easy to run a spectrogram of a record and a CD. The CD has much less discernible noise. This means quantitatively that the CD is a better, more accurate reproduction of the original signal.
Sammi is also very misled. Nyquist theorem states that a discrete waveform sampled at twice the maximum frequency of a continuous waveform can be converted perfectly into the continuous waveform via an ideal lowpass filter at half the Nyquist rate. So 44.1kHz sampling is sufficient to perfectly reproduce 20kHz tones. Unfortunately, there are no ideal low pass filters, but oversampling DSPs can do a damn-near-ideal job.
The other problem is that CDs are not merely sampled signals. They are quantized as well, so there's sowe rounding error from that process. Nonetheless, this error is less than the error [noise] introduced by analog mastering and recording. That's a mathematic fact.
I understand you're not trying to be adversarial, but you cannot get around a simple fact: Digital recording merely approximates the sound wave, no matter if it's quantized or filtered or both. It's still an estimation. Period.
However, in the real listening world, there are other factors (as you point out)...noise and susceptibility to heat, moisture, dust, warping, etc. Therefore, most people are going to hear a digital recording as being superior. Certainly the durability and convenience of it make up for any shortcomings, at least in my view.
My point is simply that a great many people feel they can hear a certain warmth and space in an excellent vinyl recording (depending on musical material, and depending on the equipment that it's played on) that is not present in a digital recording. Additionally, any digital recording has to pass through a D/A converter, which can further distort the accuracy. Ultimately, you'll hear the sound being pushed through speakers. With digital, it must first be converted. No matter how good the D/A converter is, it can't be 100% efficient.
You can try to "math your way" out it, but my money is on this: If I had you listen to a brand new analog classical recording on hi-fi equipment, then contrasted it with a CD of the same, you'd hear what I'm talking about. It's not enough to make me embrace vinyl, but it is there.
What I am interesting in doing is illustrating that a digitized recording can often be a more accurate reproduction of the original waveform that an can be an analog recording, and that in practice this is almost always the case.
That is blatantly false. It is not as accurate. It may sound better because of the lack of noise, but it's not more accurate, especially by the time it hits your ears.
To this extent I think we can all go home happy, because as far as I can tell no one is any longer in disagreement.
Digital is here to stay, so I would hope you're right.
That is blatantly false. It is not as accurate. It may sound better because of the lack of noise, but it's not more accurate, especially by the time it hits your ears.
You are wrong. Period. (as you so like to say).
Both digital and analogue recording to vinyl add noise (quantisation with dithering is exactly the same as adding random noise to the signal, that you clearly do not understand this Splinemodel and I are unable to help), vinyl adds much, much more and the digital signal is therefore more accurate.
Have you ever taken the output of a record deck, and passed it through a 16-bit/44.1 kHz AD-DA loop and compared the two pathways (i.e. "straight from record deck -> amplifier -> speaker" and "record deck->44.1 kHz/16 bit AD converter -> DA converter -> amplifier -> speaker)? I'm willing to bet you that in a properly controlled double-blind experiment you wouldn't be able to tell the difference with 95% confidence or better.
You can't take a commercial vinyl recording and commercial CD recording of allegedly the same material and use that as the comparison. CDs go through a mastering process that sadly in many cases makes a mockery of CD's enhanced dynamic range. Some folk just think "louder=better" and compress the hell out of the original signal's dynamic range (making the whole lot louder) before pressing it to CD.
In my case, I only have a CD or digital collection, so there was no nostalgia element for me. But the difference in listening experience between the CD and the vinyl was quite marked, both for me and the others in the listening room. (But there may have been some nostalgia element as all the other parties had vinyl collections as well as CDs!).
Sounds like the test wasn't double-blind. Sadly this means it had/has no validity whatsoever.
CDs being a severe compromise to start with
CD is not a severe compromise. It is perfectly adequate.
for music on regular CDs, being sampled at 44.1kHz, this means that at the higher end of the the frequency range, each cycle is only represented by very few data values (for example 10khz, thats only 4.41 samples per cycle).. such a small amount of data clearly cannot accurately represent the complex material at the top end of the frequency spectrum.
It is easy to think this, when looking on the face of it. The fact is though, that it's simply not true. As long as you have at least anything greater than two samples per period (e.g. 2.00000000001 samples is enough) of the highest frequency you want to capture, you can re-create exactly the signal you sampled. Basically, wherever your samples end up, as long as the signal conforms to the bandwidth limit criterion, there is one and only one signal that could have led to the samples being the values that they are. This can be a difficult concept to grasp.
Then there's the problem of aliasing as a result of using inadequate sample rates... see Nyquist Theorem.. To counter the (nasty) effects of aliasing, as well as smoothing out the spiky waveforms formed as a result of "not enough data", designers of digital audio equipment have included anti-aliasing digital filters into the DA conversion circuits of equipment. This is obviously far less of an issue with SuperAudioCDs since 2.8 MHz sample rates are high enough to avoid both the aliasing and "not enough data" problems and artifacts. But with ordinary CDs and (to a lesser extent) DVD audio, this is still an issue.. and the effects of digital filters on the program material, especially in cheap massproduced equipment, can be severe, especially regarding the capability of the equipment to reproduce the phase angle of the waveform, over the entire audio frequency range. Considering that we perceive directional information of a sound via the higher frequency components of sound, and these digital filters are tuned to the higher end of the frequency spectrum, then it would follow that the spatial information contained in a stereo audio recording would be an element to suffer. Just speculating here, but maybe thats why on that vinyl vs. CD demo I heard, the stereo image was so much better defined on the analog rather than the digital version.
Indeed, anti-aliasing filters can be a problem if poorly implemented.
Those who feel that CD is not adequate may be interested in this recent Audio Engineering Society Journal paper that shows that the difference between CD and higher-resolution audio formats (DVD-Audio and Super-Audio CD) is not audible. I am not aware if the paper is available for free anywhere.
You are wrong. Period. (as you so like to say).
Both digital and analogue recording to vinyl add noise (quantisation with dithering is exactly the same as adding random noise to the signal, that you clearly do not understand this Splinemodel and I are unable to help), vinyl adds much, much more and the digital signal is therefore more accurate.
No, it's not the same at all. With a vinyl recording, you get noise from the medium. With dithering, you're intentionally adding noise to compensate for inaccuracies/quantization error. You're still likely to hear a digital recording as being better because there is so much less unintentional noise. But the the analog recording is capturing the actual sound wave. This can make an audible difference at times...even if very small.
Have you ever taken the output of a record deck, and passed it through a 16-bit/44.1 kHz AD-DA loop and compared the two pathways (i.e. "straight from record deck -> amplifier -> speaker" and "record deck->44.1 kHz/16 bit AD converter -> DA converter -> amplifier -> speaker)? I'm willing to bet you that in a properly controlled double-blind experiment you wouldn't be able to tell the difference with 95% confidence or better.
Of course I haven't. Have you? It's not a valid comparison anyway. You're merely testing the D/A converter. That's not the only issue with digital.
You can't take a commercial vinyl recording and commercial CD recording of allegedly the same material and use that as the comparison. CDs go through a mastering process that sadly in many cases makes a mockery of CD's enhanced dynamic range. Some folk just think "louder=better" and compress the hell out of the original signal's dynamic range (making the whole lot louder) before pressing it to CD.
"You can't compare the gas mileage of a hybrid Toyota Camry and a non-hybrid Toytota Camry because the manufacturer limits the fuel efficiency of the hybrid for reasons X, Y and Z. Theoretically, it's way better though."
Also, I doubt you'll find many classical recordings that are compressed to all hell. I'm sure it happens a lot with pop, rock, etc.
Sounds like the test wasn't double-blind. Sadly this means it had/has no validity whatsoever.
This isn't a fucking laboratory. Sammi and I both had similar experiences. Many others have as well.
CD is not a severe compromise. It is perfectly adequate.
Absolutely, I agree. But it's not as good as many think.
It is easy to think this, when looking on the face of it. The fact is though, that it's simply not true. As long as you have at least anything greater than two samples per period (e.g. 2.00000000001 samples is enough) of the highest frequency you want to capture, you can re-create exactly the signal you sampled. Basically, wherever your samples end up, as long as the signal conforms to the bandwidth limit criterion, there is one and only one signal that could have led to the samples being the values that they are. This can be a difficult concept to grasp.
Indeed, anti-aliasing filters can be a problem if poorly implemented.
Those who feel that CD is not adequate may be interested in this recent Audio Engineering Society Journal paper that shows that the difference between CD and higher-resolution audio formats (DVD-Audio and Super-Audio CD) is not audible. I am not aware if the paper is available for free anywhere.
So your position is honestly that there is no difference between S/A-CD, DVD-Audio and CD? Why, because of this?
The authors report on a series of double-blind tests comparing the analog output of high-resolution players playing high-resolution recordings with the same signal passed through a 16-bit/44.1-kHz ?bottleneck.?.....
The test results show that the CD-quality A/D/A loop was undetectable at normal-to-loud listening levels, by any of the subjects, on any of the playback systems. The noise of the CD-quality loop was audible only at very elevated levels.
First, we're not just talking about noise. We're talking about more subjective factors...seperation, depth, warmth, tonal quality, etc.
Secondly, the noise was detectable, only at "very elevated" levels. Well what does that mean exactly? What was the source material?
Third: I don't think running the signal through a "bottleneck" is the same as listening to two different formats on the same equipment. The high resolution recording simply contains more information, particularly in terms of the sampling rate and the bit depth. I find it hard to believe that trying to "block" some of that information through a A/D/A conversion (bottleneck) would be exactly the same as listening to a source that is actually recorded differently.
No, it's not the same at all. With a vinyl recording, you get noise from the medium. With dithering, you're intentionally adding noise to compensate for inaccuracies/quantization error. You're still likely to hear a digital recording as being better because there is so much less unintentional noise. But the the analog recording is capturing the actual sound wave. This can make an audible difference at times...even if very small.
Tell me: which one of us is it that has been studying Electronics since they were 13, is now 28, has the top first in their year in a Masters Degree in Electronic Engineering and is currently doing a PhD in Audio Electronics?
This is not a matter of opinion, it is a matter of fact. I am right and you are wrong. You can say what you like, it won't change the facts. Either you can go and study this until you understand it, or you can continue to blather nonsense.
"You can't compare the gas mileage of a hybrid Toyota Camry and a non-hybrid Toytota Camry because the manufacturer limits the fuel efficiency of the hybrid for reasons X, Y and Z. Theoretically, it's way better though."
Way to prove you didn't understand the point I was making.
This isn't a fucking laboratory. Sammi and I both had similar experiences. Many others have as well.
Without performing a test double-blind, the observer-expectancy-effect/placebo heavily influences what you perceive. Someone tells you you are listening to something different makes it much more likely you will "hear" differences.
So your position is honestly that there is no difference between S/A-CD, DVD-Audio and CD? Why, because of this?
Yes. I've read the paper (and understand it). Have you? The test was well controlled and used a number of "experienced listeners". No one could hear the difference. It is probably worth mentioning at this point that I was very surprised by this. I was formerly of the opinion that some experienced listeners probably could hear the difference due to the aforementioned anti-aliasing filter problems and the improved dynamic range.
First, we're not just talking about noise. We're talking about more subjective factors...seperation, depth, warmth, tonal quality, etc.
The paper isn't just a study of the audibility of noise. The only reason noise is mentioned in the abstract is that the higher noise floor of the CD-quality A/D/A loop was audible at "very elevated" levels. The test in general was a simple comparison between the audio quality (encompassing all those things you mentioned) of high-resolution digital and CD-resolution digital. Not a single one of the listeners could hear the difference. Many tests were conducted over the course of a year, with several set-ups and many different recordings.
Secondly, the noise was detectable, only at "very elevated" levels. Well what does that mean exactly?
It means that when they turned the volume up real loud, listeners could tell when the A/D/A loop was switched in to the audio path during quiet passages, because they could hear the higher noise of that path. However, this was at volume levels that caused physical pain during the not-quiet passages.
Third: I don't think running the signal through a "bottleneck" is the same as listening to two different formats on the same equipment. The high resolution recording simply contains more information, particularly in terms of the sampling rate and the bit depth. I find it hard to believe that trying to "block" some of that information through a A/D/A conversion (bottleneck) would be exactly the same as listening to a source that is actually recorded differently.
But that's exactly what happens (or should happen) during the recording process (i.e., recording and mastering is performed at 24 bit/192 kHz, which is then converted to 16-bit/44.1 kHz, stored on CD, converted back to analogue, amplified and connected to speakers). The test shows that 16-bit/44.1 kHz is indistinguishable from 24 bit/192 kHz or SACD modulation.
The problem with CD is, as I mentioned earlier, idiots who master the recording by compressing the hell out of it. That's got nothing to do with 16-bit/44.1 kHz. It is one of the reasons that vinyl can "sound better". The actual medium though, is vastly inferior to CD in terms of signal fidelity.
Without performing a test double-blind, the observer-expectancy-effect/placebo heavily influences what you perceive. Someone tells you you are listening to something different makes it much more likely you will "hear" differences.
That experience to which I was referring was not double blind, of course. That would have been impossible, given the background noise that is always present in a vinyl recording, which gives the show away before any music starts. However, we (the audience) didn't know which medium was first up, but as soon as the music started on the vinyl version (the "B" in the A-B test), it was immediately apparent that the vinyl sounded more *real*, or *musical*. Of course the vinyl wasn't technically as perfect, given the pops, blats, clicks and surface noise from the vinyl, whereas the music from the CD emerged out of a background of digital zeros.. i.e. absolute silence (ignoring whatever noise was being generated by the amplifier, which in this case was inaudible). Yes, I know that "real","musical", "presence", "sense of space" etc etc are subjective listening terms, not scientifically quantifiable, but isn't that what listening to music is all about, and whichever equipment delivers the most "musical" reproduction is surely the most desirable? Yes, the CD has a larger dynamic range, delivers zero noise, and according to the theory (ie a perfect digital representation of the original waveform), should have beaten any vinyl competition, especially given the nature and cost of the CD playback equipment and DA converters ($ many thousands!). But that is not the impression that anybody in the audience got. I realize this is getting into dodgy ground here, and an area I am not very comfortable with, especially as a skeptic... but I know my ears fairly well, and have been to more live orchestral concerts than I can count.... but the vinyl was the clear musical winner. One can only speculate re. the cause, and that doesn't get anyone anywhere.
Regarding the claim that 44.1kHz at 16 bit resolution can deliver, in theory, audio "perfection", why has Digidesign been pushing 192kHz sample rate at 24bits for their ProTools system, and why has the audio industry even bothered developing SACD technology, when according to the stated theory, the 8 extra bits and 27,955,900 extra samples per second is "excess baggage"?
That experience to which I was referring was not double blind, of course. That would have been impossible, given the background noise that is always present in a vinyl recording, which gives the show away before any music starts.
It's not impossible to show that the CD medium itself is not a problem - you record the output of a vinyl deck, digitise it, record it to CD and do the A/B comparison that way - the noise of the vinyl is stored on the CD and then there's no noise floor difference to give the game away.
I say again that supposedly "the same" commercial recordings can be mastered completely differently for vinyl and CD - often with idiot decisions made on the CD front, resulting in better sound from vinyl. But that's not the fault of the CD medium itself.
Of course, it does boil down to if vinyl to you sounds better, it's a no-brainer that that's what you should listen to. Just don't kid yourself that it's because vinyl as a medium gives you "more" of the music, or a "more accurate" waveform.
Really, you should be getting mad at mastering folk ruining so many CD recordings - it should be giving you better sound quality but often it fails through no fault of the medium itself.
Yes, the CD … delivers zero noise
Not zero noise, just a few orders of magnitude lower than vinyl.
Regarding the claim that 44.1kHz at 16 bit resolution can deliver, in theory, audio "perfection", why has Digidesign been pushing 192kHz sample rate at 24bits for their ProTools system, and why has the audio industry even bothered developing SACD technology, when according to the stated theory, the 8 extra bits and 27,955,900 extra samples per second is "excess baggage"?
As I alluded to earlier, I don't think it was that unreasonable to expect higher-resolution mediums to deliver higher-quality audio.
In comparison to vinyl, the CD medium is superior in every way but didn't seem to quite reach the limits of human hearing in terms of dynamic range, and the anti-aliasing filters required could potentially cause issues (mentioned earlier in this thread) if not implemented properly.
Increasing bit-depth delivers higher dynamic range and increasing the sampling rate makes it possible to implement analogue anti-aliasing filters with linear phase in the entirety of audible passband. Surprisingly, given the length of time higher-res formats have been around, that paper reports on the first double-blind comparison that I'm aware of between high-res and 16-bit/44.1 kHz. As it turns out, the higher-res is a waste of time.
It's not impossible to show that the CD medium itself is not a problem - you record the output of a vinyl deck, digitise it, record it to CD and do the A/B comparison that way - the noise of the vinyl is stored on the CD and then there's no noise floor difference to give the game away.
Yes, that is correct. But given the mature of the comparison, there was no time to get into all that rigmarole.
I say again that supposedly "the same" commercial recordings can be mastered completely differently for vinyl and CD - often with idiot decisions made on the CD front, resulting in better sound from vinyl. But that's not the fault of the CD medium itself.
Agreed in principle.
So much depends on the nature of the music that is being mastered. Most modern non-orchestral material has by nature a tiny dynamic range, and is further compressed at source, in the multi-tracking, mixing and final mastering environments. If you analyze the dynamic range of (most) music that is aimed at radio play and there's barely 2 dB difference between the "quieter" and "louder" passages, at the most. When broadcast on radio, further compression is added in the controlroom... listen to some of that rare species of rock on radio that has some dynamic range, and you can hear the compression kicking in during the quieter passages... sometimes to the extent that noise encoded in older recordings, (or even the noise generated by the playback equipment in the controlroom!) is pumped up by the compressors as they attempt to keep the audio at a "flatline" level. Since most music is listened to casually as part of the furniture (as opposed to active listening), and often in unfriendly listening environments such as in vehicles, and other places with a high ambient noise level (or other distractions), quieter passages will be lost, or even inaudible.
Orchestral music however has a dynamic range (from 'ppp' to 'fff') of up to 75dB, and the relative difference in SPL is an essential component of the music, an essential compositional device. Different mastering techniques are employed for "classical" music. But for classical music radio broadcast it is a necessary compromise to employ some degree of compression otherwise the quieter passages will be lost to most listeners.. of course, classical recordings cover so much different varieties of music with different amounts of dynamic range, and some will suffer far more that others. However, for those who actively listen to classical music on quality audio playback systems at home, the ability to reproduce the full dynamic range is essential for the listening experience.
And yes, CD, re. ability to reproduce dynamic range, is superior to vinyl or tape.
Of course, it does boil down to if vinyl to you sounds better, it's a no-brainer that that's what you should listen to. Just don't kid yourself that it's because vinyl as a medium gives you "more" of the music, or a "more accurate" waveform.
Oh agreed. But I am still curious as to the "why" of vinyl, or even better, half-inch analog tape at 30ips. Its a bit like state of the art tube amplifiers "holding their ground" compared to state of the art transistor amplifiers, yet the quoted technical specifications of such tube amps are (as a rule) vastly inferior to their transistorized counterparts....
Really, you should be getting mad at mastering folk ruining so many CD recordings - it should be giving you better sound quality but often it fails through no fault of the medium itself.
There are plenty of cowboys out there (really), in amongst a small minority of highly skilled mastering engineers. And there are many recordings out there which sound "brittle and edgy" (almost unlistenable for me), especially amongst recordings that used early digital equipment, from the late 1970s and early 1980s.
As I alluded to earlier, I don't think it was that unreasonable to expect higher-resolution mediums to deliver higher-quality audio.
In comparison to vinyl, the CD medium is superior in every way but didn't seem to quite reach the limits of human hearing in terms of dynamic range, and the anti-aliasing filters required could potentially cause issues (mentioned earlier in this thread) if not implemented properly.
Increasing bit-depth delivers higher dynamic range and increasing the sampling rate makes it possible to implement analogue anti-aliasing filters with linear phase in the entirety of audible passband. Surprisingly, given the length of time higher-res formats have been around, that paper reports on the first double-blind comparison that I'm aware of between high-res and 16-bit/44.1 kHz. As it turns out, the higher-res is a waste of time.
I understand what you are saying, but are you now implying that 44.1/16[I] is a compromise, re. the need for a "perfect" brickwall AA filter to allow for an exact digital copy of the waveform at the 44.1kHz sample-rate (which is the absolute minimum needed to reproduce the full audio spectrum) and bit depth?
Regarding the claim that 44.1kHz at 16 bit resolution can deliver, in theory, audio "perfection", why has Digidesign been pushing 192kHz sample rate at 24bits for their ProTools system, and why has the audio industry even bothered developing SACD technology, when according to the stated theory, the 8 extra bits and 27,955,900 extra samples per second is "excess baggage"?
I've been out of the audio recording loop for nearly a decade but here's my 2 cents.
1) It used to be conventional thinking that when you start mixing tens to hundreds of 16 bit/44.1 tracks together you need greater bit precision - not ultimately for the dynamic range of the music, but because of the mathematics involved in adding what is essentially a collection of 1's and 0. For a long time 24 bit devices, while theoretically better, had worse recording/reproduction characteristics than 16/44.1 - maybe thats why the ultimate pro gear didn't move to 24 bits right away.
It is also beneficial to have headroom when recording because digital clipping (excess signal over the available bandwidth) is quite nasty compared to analog clipping - which is quite nice - and thats part of the reason vinyl has a distinctive character - its analog clipping - which is a form of compression in itself as it is non-linear and harmonic, while digital clipping is absolute, harsh and disharmonic. Headroom is very important, because naturally played instruments aren't compressed by default when recording digitally, so anything over 0db will cause nasty distortion. When recording analog, signals over 0db are non-linearly attenuated because of the magnetic saturation of the medium, which is like I say just a 'nice' kind of distortion.
Even then you still take your master mix of several tens of 24/192 tracks and dither the final output to 16 bits. This isn't done because it sounds better per se, but because of the maths i mentioned. All you are trying to do is create a 2 channel mix, so you need to do the maths at 24/192 because you have lots of individual tracks - but the final 2 channel is as perfectly good at 16/44.1 or 48. This doesn't really change the waveform of the finished music*, but just scales it to fit in a 16 bit resolution. I dont know if it is still the case, but early systems used to just drop the few least significant bits to scale to 16 bit - because you simply cannot hear them - i suspect these days its a bit more sophisticated.
*i know it does theoretically, but due to nyquists theorem and a half decent filter you cannot tell the difference. Back in the day I remember reading in the journal Sound on Sound - I had a pile of them from floor to ceiling - hehe! that research was ongoing into the psychoacoustic effect of frequencies above the threshold of human hearing, ie 20khz being percievable - not as sound, but in the effects it has on sound in the range of hearing. I dont know what the outcome of this was - and I remember Pioneer releasing cd players that interpolated higher than human range frequencies - up to about 60khz IIRC, but i dont think that ever caught on.
Microphones do not have the range that 24bits can store - and if you look at the most expensive mics in the world - it really isn't the quality of reproduction of the source that makes them sought after - but the analog clipping and shaping they perform. Tubes in particular sound lovely - but thats because they are non-linear compressors, saturators and distorters in effect.
It used to be the case, Im not sure if it still is, that even though the ADC's were rated at 24 bits, the actual electronics involved could not actually perform at 24 bits anyway. The SNR ratio is so high - IIRC something like 132db that you're actually hitting the limits of the theoretical maximum attainable SNR which is ultimately limited by the random noise of molecular motion as governed by quantum physics.
2) Every computerised system i've ever come across suffers from what I would call the marketing BS factor. Ignorant people like big numbers. Thats why they rate monitors at 5000:1 contrast, 700 cd/m3 - even though you'd have no retinas left if you used it like that. Its why DDR3 ram is rated at 2ghz though it performs barely 5% better than 667 mhz DDR2. Its why people think 16x PCIe performs twice as good as 8xPCIe for graphics when the difference is less than 1%
'The industry' largely develops things because its something new to sell you at a greater margin. Now that CD players can be had for $10 and CD sales are declining, pushing a new format and having the public re-invest their catalog in the new format is a money spinner. 24/192 is not 4x better than CD, maybe it is 5% better for the 1% of people who can tell the difference, but cue the law of diminishing returns...its not because of the quality at all.
Its all just shit. Mr H is spot on. There is a quality about vinyl that makes it attractive, but thats because of what it 'adds' to the recorded wave in terms of analog clipping, saturation and distortion, not because it reproduces anywhere as good as CD.
If you want to hear something truly amazing, get hold of an SPL vitalizer and run your CD's and vinyl through that. It gets tiring on the ears and was much overused - but just as an exercise its good fun.
Another thing - vinyl is absolutely rubbish for off centre deep bass sounds - all kinds of phase distortions from the stylus not being able to accurately follow the grooves in the disc
But the the analog recording is capturing the actual sound wave. This can make an audible difference at times...even if very small.
Im afraid that is a popular misconception. While we all know the limitations of digital recording, to say that analog captures the actual sound wave is completely false.
Analog circutry and recording has all of the things built in by default that you deride against digital recordings. Compression, saturation, distortion. its unavoidable by the laws of physics. Much research has gone into reproducing the characteristics of analog limitations and distortions so that we can add it to digital recordings to make it sound more pleasing. If in some strange parallel universe digital had come before analog, then analog would be totally slated as an accurate recording medium,
It may sound more pleasing - but thats 180 degrees to the complaint. Digital has its own characteristics of compression, saturation and distortion, but the reason you like analog is the way it distorts!
The only way to hear what the musician intended is to hear it live - but thats not really the case because the venue will add its own form of wave-shaping to the sound of the instruments.
Thats what I was referring to re. CDs being a severe compromise to start with (that is before they are reduced yet further by one of compression algorithms to Mpeg3 formats!). Considering that the human audio frequency range is (approx.) 20hz - 20khz -- for music on regular CDs, being sampled at 44.1kHz, this means that at the higher end of the the frequency range, each cycle is only represented by very few data values (for example 10khz, thats only 4.41 samples per cycle).. such a small amount of data clearly cannot accurately represent the complex material at the top end of the frequency spectrum. design... as if that section of the device is an afterthought... ask any serious audio designer. Unless one is paying top dollar, one can guarantee that the analog audio components and circuit design is as cheap as the manufacturer can get away with. Even some "top end" digital studio consoles use opamps and associated circuitry with could never be honestly described as truly "high fidelity". So, when it comes to the "weakest link in the chain" which determines the audio quality for the listener, there are so many places where the signal is compromised. "SuperAudioCDs" will only reproduce what is there in the first place, and if the program material is sonically compromised, then that compromise (error) will be more accurately reproduced as well!
'Scuse the length of this reply!
I dont think its valid to say...20 years ago I compared this to that.
20 years ago digital recording was in its infancy, while vinyl was at a near prime. The spec for CD's hasn't changed, but the technology for getting those bits onto the disc was rather crude.
Its a bit like comparing DVD to Blu-ray. Yes the first gen of blu-ray might take 30 seconds to display the root menu, and skip a few frames, sluggish fast-forward etc, but then the computational power to do this is several magnitudes higher than DVD. Give it a few years for Blu-Ray players to attain the power and all the faults of the new medium go away.
But the potential for blu-ray is and always was set right from the beginning.
Also youre complaining that the cheap end of digital devices have crap filters etc - well the cheap end of analog is particularly crap if I recall - far worse than the crap end of digital reproduction.
The other thing when critisizing distortion -remember that distortion creates harmonics. By default a harmonic is a multiple of the fundamental. So even if you're distorting a 20khz wave - something I doubt you can hear anyway - the first harmonic of this is going to be 40khz - something you definately wont hear. While I agree that the top end of audio is compromised by recording digitally, when done properly, it isn't something to worry over. Also high frequencies (and low frequencies) aren't 'heard' in the same way as mid-ranges. Its quite easy to determine the timbre of a sound in the mid ranges, but when it comes to high-frequencies (and low), what is detected is more of a 'presence of energy' than a timbre. Therefore even mildly distorted high frequency waves are not detectable by the ear to be distorted - and the brain has alot to do with this too - because you are only sensing presence of energy, rather than timbre. IIRC the switch from timbre to 'presence' occurs quite low, at about 5khz, and bear in mind that any frequency you hear above say 5khz is likely to be a high-order harmonic created from fundamentals around 100hz to 3khz on average, the amount of timbral distortion you will perceive above 10khz is going to be near negligable.
Tell me: which one of us is it that has been studying Electronics since they were 13, is now 28, has the top first in their year in a Masters Degree in Electronic Engineering and is currently doing a PhD in Audio Electronics?
This is not a matter of opinion, it is a matter of fact. I am right and you are wrong. You can say what you like, it won't change the facts. Either you can go and study this until you understand it, or you can continue to blather nonsense.
Way to prove you didn't understand the point I was making.
Way to be an ass. If you can't discuss the topic at hand without being condescending and rude, then go away.
As for not understanding your point, you're wrong. I think it's you who doesn't understand the simple point that a CDs theoretical performance does not equate to a CDs typical performance. Are there idiots mastering CDs? Absolutely. According to you, they may all be idiots. The fact is though...it doesn't matter. We can buy what we can buy.
Without performing a test double-blind, the observer-expectancy-effect/placebo heavily influences what you perceive. Someone tells you you are listening to something different makes it much more likely you will "hear" differences.
I didn't know what I was listening to or that it was different. In fact, I had assumed it was a CD.
Yes. I've read the paper (and understand it). Have you? The test was well controlled and used a number of "experienced listeners". No one could hear the difference. It is probably worth mentioning at this point that I was very surprised by this. I was formerly of the opinion that some experienced listeners probably could hear the difference due to the aforementioned anti-aliasing filter problems and the improved dynamic range.
So you're basing your position on a single study. Awesome.
The paper isn't just a study of the audibility of noise. The only reason noise is mentioned in the abstract is that the higher noise floor of the CD-quality A/D/A loop was audible at "very elevated" levels. The test in general was a simple comparison between the audio quality (encompassing all those things you mentioned) of high-resolution digital and CD-resolution digital. Not a single one of the listeners could hear the difference. Many tests were conducted over the course of a year, with several set-ups and many different recordings.
Yes, but was not a test of actual recordings played on typically used equipment. It was a test of the loop's effect, and nothing more. A more effective and realistic test would be to compare two recordings in different formats, because that's what people are able to purchase anyway. I'm sure some recordings would show more of a difference than others because of the differences in mastering. Even if the differences heard are in the mastering, those differences are still there.
It means that when they turned the volume up real loud, listeners could tell when the A/D/A loop was switched in to the audio path during quiet passages, because they could hear the higher noise of that path. However, this was at volume levels that caused physical pain during the not-quiet passages.
But the difference was there.
But that's exactly what happens (or should happen) during the recording process (i.e., recording and mastering is performed at 24 bit/192 kHz, which is then converted to 16-bit/44.1 kHz, stored on CD, converted back to analogue, amplified and connected to speakers). The test shows that 16-bit/44.1 kHz is indistinguishable from 24 bit/192 kHz or SACD modulation.
Is all or the majority of recording actually done at 24 bit/192KHZ? Also, what about recording at much higher sampling rates, such as 2.8MHZ?
The problem with CD is, as I mentioned earlier, idiots who master the recording by compressing the hell out of it. That's got nothing to do with 16-bit/44.1 kHz. It is one of the reasons that vinyl can "sound better". The actual medium though, is vastly inferior to CD in terms of signal fidelity.
Even if so, it is what it is. And once again, I don't know that vinyl "sounds better" overall. I simply know that there are many, many people who've had the same experiences sammi and I have. They listened to something and thought "wow, that's got such warmth and depth and space to it....what a great CD." Then they found out it was vinyl.
.....
Also high frequencies (and low frequencies) aren't 'heard' in the same way as mid-ranges. Its quite easy to determine the timbre of a sound in the mid ranges, but when it comes to high-frequencies (and low), what is detected is more of a 'presence of energy' than a timbre. Therefore even mildly distorted high frequency waves are not detectable by the ear to be distorted - and the brain has alot to do with this too - because you are only sensing presence of energy, rather than timbre. IIRC the switch from timbre to 'presence' occurs quite low, at about 5khz, and bear in mind that any frequency you hear above say 5khz is likely to be a high-order harmonic created from fundamentals around 100hz to 3khz on average, the amount of timbral distortion you will perceive above 10khz is going to be near negligable.
But those "energies" are perceived, and may well result in our perceptions of space, warmth, depth, etc.
Im afraid that is a popular misconception. While we all know the limitations of digital recording, to say that analog captures the actual sound wave is completely false.
Uh, OK.
Analog circutry and recording has all of the things built in by default that you deride against digital recordings.
I'm not deriding anything.
Compression, saturation, distortion. its unavoidable by the laws of physics.
I agree.
Much research has gone into reproducing the characteristics of analog limitations and distortions so that we can add it to digital recordings to make it sound more pleasing.
If in some strange parallel universe digital had come before analog, then analog would be totally slated as an accurate recording medium,
Yes, but we are adding those things, which simply cannot be the same as them existing naturally. In fact, I believe it was you who stated that wrt clipping in an earlier post.
It may sound more pleasing - but thats 180 degrees to the complaint. Digital has its own characteristics of compression, saturation and distortion, but the reason you like analog is the way it distorts!
First, I don't prefer analog overall (for about the 7th time now, actually). Secondly, even if one prefers analog because of the natural distortion, it's still a difference that's worth noting. It's not the same as digital.
The only way to hear what the musician intended is to hear it live - but thats not really the case because the venue will add its own form of wave-shaping to the sound of the instruments.
Obviously.
Lets address another thing - the presence of energy that you perceive rather than hear. First off I dont fully know the answer to this question, however I think it is possible to look at other systems that I do know about to get a good guesstimate about how likely it is that ultrasonic waveforms have a perceptible effect on normal sounds.
A quick look on Wiki confirms that ultrasonic noises, those above 20khz can have an 'impression' on the mind - if the sound is 'wired' directly to the skull and bypasses the bones in the ear.
http://en.wikipedia.org/wiki/Ultrasound
The question is - is it this effect that is responsible for the pleasing characterists of vinyl vs digital? I think the answer is no.
Lets go back a step and consider the sampling of every individual note of a piano at 24/196 for a massive multisampled instrument to get a clear picture of the variables at play.
The problem with sampling every note of a piano is one of resonances between the strings. In effect all other things being equal, striking one note causes other strings to vibrate in sympathy, but only certain other ones. The way this is worked out is that if a string has an overtone in common of one of the harmonics, a resonant wave will be set up.
So if you pluck a string say 200hz, then strings tuned to 400hz, 600hz, 800hz etc will resonate. However, you have to look at the waveform of the fundamental to see how much energy is transfered to the other strings. Assuming a sawtooth wave - which has all odd and even harmonics falling away exponentially in energy from the fundamental, you can visualize that it would take one hell of a 200hz wave to get any meaningful resonances past the 10th harmonic.
That being said, after you have sampled every note of the piano, (which incidently does not look anything like a sawtooth wave, it has far fewer harmonics, you might think you have the perfect sounding piano - all resonances included) . Nope!
The voodoo thing ... is when you strike 2 notes together on a real piano - it sets up a different set of resonant interplay between the strings than you would verifyably get if you hit both notes seperately. And offcourse you have not sampled these variations...You cant accurately sample a piano - unless you take a sample of every single combination of the 88 keys - and that is not possible.
However - we have looked at this in reverse of what we are trying to find out. We have been hitting a low note and finding out if higher tuned strings resonate. They do. Now what happens if you hit a high note? do lower tuned strings resonate?
They do....But to what extent?
But its more complex. In order for a lower note to resonate in sympathy with a higher note, it has to have a harmonic sequence that at some point coincides with the fundamental or one of the harmonics of the high note.
So for instance, it I strike a virtual 1000hz sawtooth, my first harmonic is going to be 2000hz - but at half the energy. Theoretically, a 500hz virtual string will have a 3rd harmonic at 2000hz so they can resonate, however the 3rd harmonic of a sawtooth is 1/3 the energy of the fundamental, so even in the best theoretical example you can imagine, the transfer of resonant energy is pretty miniscule. Especially when you consider that sound is measured on a logarathmic scale BUT heard only as one continuous wave by the ear - which is deconstructed in the brain - which is really voodoo.
But lets take this too our digital ultrasonic example. Can say a 40khz wave cause resonances in audible sounds that to any practical extent can be perceived by a human - either consciously or subconsciously.
Starting with the basics. Any wave that is created at 40khz is going to be by default a very high order harmonic of something. That means it contains very little energy. Sawtooths are extreme examples - most instruments produce nowhere near as much harmonics as a sawtooth - or square wave.
Any ultrasonic harmonic produced by a natural instrument is going to be extremely low energy. Its potential for causing a resonance in the audible spectrum is therefore extremely low. Then factor in that an audible sound has to have an harmonic in common with the ultrasonic sound to resonate, and again by default, this harmonic is going to be of a very high order with a miniscule amount of energy attached to it. To all extents and purposes the resulting cross product resonance is basically nothing.
Now if you take a high frequency pure-tone audible wave in complete isolation and resonate it with a single tone ultrasound and measure the distortion very accurately - in labratory settings - you will find a very small detectable resonance - and this is what I believe the hi-fi freaks have latched on to. They're shouting "look its there - ITS THERE!!!' -Im missing something yadda yadda yadda!!!"
But its not there, it was created by a special electronic circuit designed to output things you cannot hear - and more importantly - really would not want to hear.
What they're missing apart from a few marbles, is that this is not the realworld. There is no music id want to listen too that contained nothing but 20khz puretones modulated with ultrasonic frequencies - and that the whole test is designed to show what happens in the very worst case scenario - again something that never happens in the realworld.
Then factor in that even in the very best recording studios, by the time the ultrasonic wave has passed through all the equipment that is needed to record and duplicate the sound in the first place - that is IF the microphone even picked it up in the first place.....It simply does not exist in the vinyl recording anyway!
It is not the Low Pass Filters, Bit rate and sampling frequency (ie lack of ultrasound) that are the issues in recordings made to CD's - just as it is not the presence of ultrasonic sound that makes vinyl sound the way it does.
The pleasurable warmth of vinyl recording is because of magnetic saturation, non-linear distortion, magnetic hysterisis, eddy currents, frequency responce of capacitors and inductors in the circuit....and a million other things - ALL OF WHICH SERVE TO DEVIATE THE RECORDED WAVEFORM FROM THE ORIGINAL- yes it sounds nice, but thats the black art form of the audio designer.
From a technical, accuracy, perspective, when done right, CD's reproduce the original waveform much more accurately, subject to thier characteristic distortions.
The funniest thing is, what it comes out of, is and will always be an analog device. The Loudspeaker. When you physically reproduce the sound on your loudspeaker, the cone has to move between the positions of any two samples anyway, thus smoothing the sampled waveform of its sampling frequency, bit depth and step distortion- thus interpolating all the missing *theoretical* perceived energy you cant hear anyway - anyway!!!
Yes, but we are adding those things, which simply cannot be the same as them existing naturally.
i'll address this point in particular.
What they're doing is analysing the reproductive characteristics of analog devices using probably not much more than simple sine waves, measuring how the analog circuitry distorts the wave over the entire audible spectrum, creating a computer model of frequency response and adding back in the same kinds of distortion.
So which manufacturers model of analog circuitry do you analyse and model? What turntable maker sounds the absolute best??? very subjective....
It will never sound exactly the same as vinyl. However there is a point that will be reached where the decibel level of missing distortion or error is so small as to be indistinguishable.
However - consider this thought experiment.
If vinyl sounds so good, why would the clued-up audio engineers simply not record their best original source to vinyl and sample the vinyl for release on CD?
What would happen if you did it vice-versa.
Which would produce the most accurate reproduction of the original sound as the musician intended?
First a dislaimer. I have not studied this for a long time, I cant promise i've outlined every principle 100% accurately.
Secondly, I dont really care either way in the analog vs digital debate, a good system either analog or digital can sound fabulous.
Thirdly, if you prefer either principle - why cant it just be 'because I/you do'. From what I can fathom, most audiophiles completely invent the most ridiculous psuedosciencejunk for their reasoning.
Fourth - lets not get started on 99.9% oxygen free cabling...
well SDW, I was addressing things in the sense of the general complaint about analog vs digital - if you're fine with it good on you!
I am fine with it. As I've said, digital wins hands down for me. The advantages are enormous. All I'm saying is that many people analog recordings have a certain warmth, depth, space (etc.) that digital ones don't have...and I tend to agree with them, at times.
Lets address another thing - the presence of energy that you perceive rather than hear. First off I dont fully know the answer to this question, however I think it is possible to look at other systems that I do know about to get a good guesstimate about how likely it is that ultrasonic waveforms have a perceptible effect on normal sounds.
A quick look on Wiki confirms that ultrasonic noises, those above 20khz can have an 'impression' on the mind - if the sound is 'wired' directly to the skull and bypasses the bones in the ear.
http://en.wikipedia.org/wiki/Ultrasound
The question is - is it this effect that is responsible for the pleasing characterists of vinyl vs digital? I think the answer is no.
I don't know...I was just throwing it out there. It might, it might not. I just think there has to be an explanation. I find it unlikely that all the people out there who hear analog as being different/better are victims of a placebo or nostalgia effect.
Lets go back a step and consider the sampling of every individual note of a piano at 24/196 for a massive multisampled instrument to get a clear picture of the variables at play.
The problem with sampling every note of a piano is one of resonances between the strings. In effect all other things being equal, striking one note causes other strings to vibrate in sympathy, but only certain other ones. The way this is worked out is that if a string has an overtone in common of one of the harmonics, a resonant wave will be set up.
That's much more complex than you've made it out to be. First, the piano is dampened. Secondly, the overtones in the string are going to be present regardless of any other sympathetic vibrations.
So if you pluck a string say 200hz, then strings tuned to 400hz, 600hz, 800hz etc will resonate. However, you have to look at the waveform of the fundamental to see how much energy is transfered to the other strings. Assuming a sawtooth wave - which has all odd and even harmonics falling away exponentially in energy from the fundamental, you can visualize that it would take one hell of a 200hz wave to get any meaningful resonances past the 10th harmonic.
OK....
That being said, after you have sampled every note of the piano, (which incidently does not look anything like a sawtooth wave, it has far fewer harmonics, you might think you have the perfect sounding piano - all resonances included) . Nope!
Actually, that's a poor example because pianos are tuned under Equal Tone Temperament (mathematically imperfect, but equal intervals). So sampling every note is fine, but the notes are not mathematically in tune.
The voodoo thing ... is when you strike 2 notes together on a real piano - it sets up a different set of resonant interplay between the strings than you would verifyably get if you hit both notes seperately.
I'm not sure that's true. Where did you get that from?
And offcourse you have not sampled these variations...You cant accurately sample a piano - unless you take a sample of every single combination of the 88 keys - and that is not possible.
There is no need for that because every interval is equal but imperfect. All one needs to do is sample each of the 88 pitches. The combinations don't matter. What you say would be true (I think) if the piano was tuned with Just Intonation or another system.
However - we have looked at this in reverse of what we are trying to find out. We have been hitting a low note and finding out if higher tuned strings resonate. They do. Now what happens if you hit a high note? do lower tuned strings resonate?
They do....But to what extent?
But its more complex. In order for a lower note to resonate in sympathy with a higher note, it has to have a harmonic sequence that at some point coincides with the fundamental or one of the harmonics of the high note.
So for instance, it I strike a virtual 1000hz sawtooth, my first harmonic is going to be 2000hz - but at half the energy. Theoretically, a 500hz virtual string will have a 3rd harmonic at 2000hz so they can resonate, however the 3rd harmonic of a sawtooth is 1/3 the energy of the fundamental, so even in the best theoretical example you can imagine, the transfer of resonant energy is pretty miniscule. Especially when you consider that sound is measured on a logarathmic scale BUT heard only as one continuous wave by the ear - which is deconstructed in the brain - which is really voodoo.
But lets take this too our digital ultrasonic example. Can say a 40khz wave cause resonances in audible sounds that to any practical extent can be perceived by a human - either consciously or subconsciously.
Starting with the basics. Any wave that is created at 40khz is going to be by default a very high order harmonic of something. That means it contains very little energy. Sawtooths are extreme examples - most instruments produce nowhere near as much harmonics as a sawtooth - or square wave.
Any ultrasonic harmonic produced by a natural instrument is going to be extremely low energy. Its potential for causing a resonance in the audible spectrum is therefore extremely low. Then factor in that an audible sound has to have an harmonic in common with the ultrasonic sound to resonate, and again by default, this harmonic is going to be of a very high order with a miniscule amount of energy attached to it. To all extents and purposes the resulting cross product resonance is basically nothing.
Now if you take a high frequency pure-tone audible wave in complete isolation and resonate it with a single tone ultrasound and measure the distortion very accurately - in labratory settings - you will find a very small detectable resonance - and this is what I believe the hi-fi freaks have latched on to. They're shouting "look its there - ITS THERE!!!' -Im missing something yadda yadda yadda!!!"
But its not there, it was created by a special electronic circuit designed to output things you cannot hear - and more importantly - really would not want to hear.
What they're missing apart from a few marbles, is that this is not the realworld. There is no music id want to listen too that contained nothing but 20khz puretones modulated with ultrasonic frequencies - and that the whole test is designed to show what happens in the very worst case scenario - again something that never happens in the realworld.
Then factor in that even in the very best recording studios, by the time the ultrasonic wave has passed through all the equipment that is needed to record and duplicate the sound in the first place - that is IF the microphone even picked it up in the first place.....It simply does not exist in the vinyl recording anyway!
I don't know. It seems to me that the analog recording (or super high quality digital recording) would be better able to include those "inaudible" tones. They may or may not have an effect on what is "heard" or even felt/perceived. It seems like one possible explanation.
It is not the Low Pass Filters, Bit rate and sampling frequency (ie lack of ultrasound) that are the issues in recordings made to CD's - just as it is not the presence of ultrasonic sound that makes vinyl sound the way it does.
I don't entirely agree, unless you're saying that while say, an 8 bit recording is not good, a 16 bit recording will always be capable of performing at a level beyond our ability to hear.
The pleasurable warmth of vinyl recording is because of magnetic saturation, non-linear distortion, magnetic hysterisis, eddy currents, frequency responce of capacitors and inductors in the circuit....and a million other things - ALL OF WHICH SERVE TO DEVIATE THE RECORDED WAVEFORM FROM THE ORIGINAL- yes it sounds nice, but thats the black art form of the audio designer.
All contributing factors to be sure, but not the only ones.
From a technical, accuracy, perspective, when done right, CD's reproduce the original waveform much more accurately, subject to thier characteristic distortions.
That's where I disagree. It's simply not true. The wave is only sampled and reconstructed. Not much is lost or estimated incorrectly, but something is. The problem in the analog comes from noise and distortions in the analog recording, which are always present. The only question is in terms of degree. High quality digital (CD and better) is so good that the vast majority of people will judge it as being "more accurate" and certainly "better" than analog. I'm certainly in that camp.
The funniest thing is, what it comes out of, is and will always be an analog device. The Loudspeaker. When you physically reproduce the sound on your loudspeaker, the cone has to move between the positions of any two samples anyway, thus smoothing the sampled waveform of its sampling frequency, bit depth and step distortion- thus interpolating all the missing *theoretical* perceived energy you cant hear anyway - anyway!!!
That's also what's amazing about people that buy electronics (amazingly stupid, that is). They'll go buy a $400 receiver and a $150 pair of speakers and think they've bought something good. In reality they've gotten one step away from the worst of the worst. In my case, I'm looking at spending $1500 for a receiver and another $1000 for speakers...and I know I'll still only have a "mid-fi" product at best. Really "good" speakers start at about $5000 a pair.
i'll address this point in particular.
What they're doing is analysing the reproductive characteristics of analog devices using probably not much more than simple sine waves, measuring how the analog circuitry distorts the wave over the entire audible spectrum, creating a computer model of frequency response and adding back in the same kinds of distortion.
So which manufacturers model of analog circuitry do you analyse and model? What turntable maker sounds the absolute best??? very subjective....
It will never sound exactly the same as vinyl. However there is a point that will be reached where the decibel level of missing distortion or error is so small as to be indistinguishable.
However - consider this thought experiment.
If vinyl sounds so good, why would the clued-up audio engineers simply not record their best original source to vinyl and sample the vinyl for release on CD?
What would happen if you did it vice-versa.
Which would produce the most accurate reproduction of the original sound as the musician intended?
Of the two...CD or whatever digital format one is using is going to be 'better." There's no question. I was just trying to point out a possible explanation for the differences heard in analog v. digital.
I don't know...I was just throwing it out there. It might, it might not. I just think there has to be an explanation. I find it unlikely that all the people out there who hear analog as being different/better are victims of a placebo or nostalgia effect.
Its very real, analog sounds different. Whether its 'better' is subjective dont you agree? Im perfectly fine with people saying they prefer the sound of analog, because what they prefer is up to them. However their reasoning is rubblish.
They prefer the sound of analog because of the way the circuitry mangles the sound!!!!
That's much more complex than you've made it out to be. First, the piano is dampened. Secondly, the overtones in the string are going to be present regardless of any other sympathetic vibrations.
well of course its much more complex than I said, but im not going to write 20 pages....However the fundamental principles are sound. (sic) hehe. Every instrument bar a sinewave generator has harmonics and overtones, thats what gives a sound its timbre, but you already know this being a muzak teacher....
Actually, that's a poor example because pianos are tuned under Equal Tone Temperament (mathematically imperfect, but equal intervals). So sampling every note is fine, but the notes are not mathematically in tune.
I'm not sure that's true. Where did you get that from?
There is no need for that because every interval is equal but imperfect. All one needs to do is sample each of the 88 pitches. The combinations don't matter. What you say would be true (I think) if the piano was tuned with Just Intonation or another system.
well, getting technical, its because a piano is not tuned mathmatically perfectly that makes it a pain to sample, the resonances are close, but the effect of imperfect intervals is that it creates a chorus effect, ie periodic modulated slight detuning.
ok, im just trying to explain things simply....vibration of strings, resonances, how they work, forgetting the specific mechanics of a piano, this is how it works - and as far as were concerned - trying to establish if ultrasound effects audio, then the piano string model is easy to visualize....
However - I am perfectly right when I say that you cannot accurately capture the subtle resonances of a piano when sampling it a note at a time. Here's why
If I strike a single note - I set the string vibrating, and on its own it contains its characteristic set of harmonics that define it as a piano - However part of the characteristic of a piano is the resonance that occurs when a string is struck that sets other strings vibrating, regardless of whether they are mathmatically in tune....harmonics tend to wander from mathmatical perfection as you hit higher order ones anyway.
Now say I sample a single key - lets "symbolically" call this A, and suppose strings B, C, D, E resonate with it. I record this
A---BCDE
Now I sample another single key lets call this A1---and strings B1,C1,D1 and E1 resonate with it.
Now in my sampler, I play A and A1 together - and as i've captured the resonance of these strings, Im also reproducing resonances B,B1,C,C1,D,D1 and E and E1 as you would expect.
A perfect capture of the essence of a piano??? Not quite.....
In a real piano, the soundwaves of A and A1 would modulate eachother - something like the effect of beating on closely matched tones, and another frequency and harmonic set is created, A2---B2,C2,D2,E2 which in turn causes other resonances in strings, and thus we have in effect a characteristic of the piano that has not been captured by our sampler when we sampled every key...because playing back the 2 samples simultaneously does not recreate A2,B2,C2,D2,E2.
Lets forget the effect on timbre and harmonics by hitting the keys harder or softer...
Now, not much music is simply 2 keys of a piano, so imagine the very complex interplay between numerous notes being played and the resonances created in the physical world..its just not possible to sample it one key at a time.
However, like the ultrasound effect we are trying to solve, the effect of this is so small and virtually indistinguishable, that you can sample every key of a piano and fool 99.9% of the listeners...if you are very clever and know how to process the sound to get around the pitfalls of sampling a piano. Its really not as easy as sampling every key.
I don't know. It seems to me that the analog recording (or super high quality digital recording) would be better able to include those "inaudible" tones. They may or may not have an effect on what is "heard" or even felt/perceived. It seems like one possible explanation.
well it cant hurt to have them present can it?
That's where I disagree. It's simply not true. The wave is only sampled and reconstructed. Not much is lost or estimated incorrectly, but something is. The problem in the analog comes from noise and distortions in the analog recording, which are always present. The only question is in terms of degree. High quality digital (CD and better) is so good that the vast majority of people will judge it as being "more accurate" and certainly "better" than analog. I'm certainly in that camp.
The 'something' that is being lost is actually - what is being added to the analog recording that makes it perceptibly sound better.
The real problem with digital is that it suffers from being too accurate and the slight distortions it does introduce are not pleasant sounding to the ear. In contrast analog is not very faithful to the source, but includes by default distortions that are pleasant to the ear.
Really, if you dont believe me that distortions make things sound better, get an SPL vitalizer or suchlike, its a psychoacoustic processor, It distorts the crap out of sounds and they take on all the properties that people are claiming make analog sound better. Better stereo seperation, high frequency distortion to add in near ultrasonic harmonics...etc. It really isn't accurate to source in any sense of the word, but it works.