Hi. Googling on the subject, I came to this site. Found it is an interesting site in the process ;-)
Forgive me, if my English seems a 'bit off'. I'm from Denmark.
On subject: 'Fact' are _not_ irrelevant. Fact is, that what the artists produced at the recording is most precisely recorded on vinyl as opposed to CD. No debating that. It's logic and physics.
I read an article, that most young people today have never even _heard_ analog music, not to mention on anything else than, say, an IPod.
Not a real stereo with 'real' load speakers.
It's the generation of low quality MP3, which is why, I suspect, the vinyl has had a renaissance of sorts. We old dudes (44 ;-) have grown tired of low quality.
An anecdote from a real experience:
In 1982 I bought a Luxman LV-103U hybrid tube amplifier and it's CD player match, also with tubes. IT WAS expensive and the CD player was top of the line(s) at that time.
I love classical music (not only ;-) and I thought this would out match my Thorens 318 turn table any day.
It did NOT.
The CD player had a very specific problem with violins.
The problem is a bit difficult to explain, but it was extremely clear and impossible to not notice:
Whenever violin-players 'started setting the bow to the string' (make any sense ?), the result coming from the speakers was the sound of an ice skater braking. Remember having seen/heard on TV, a figure-skater stopping from 'high-speed' and stopping 'side-ways'?
And I could repeat the effect with virtually any CD I had on the shelve.
As a result, and very dissapointed, I went back to the store (KT Radio in Odense, Denmark) with all my equipment. We tried at least four different high end CD players in different configureations, and they all showed the exact same problems with different CD's and recordings.
Props though to KT Radio: They actually gave me a new OM 30 Ortofon needle for my Thorens, for my trouble ;-)
Customer satisfaction pure!! :-)
The "facts" are irrelevant: you cannot prove something that is subjective.
Digital sounds slightly clearer, has the benefit you can listen to a whole album without getting up and is easier to rip to your computer; [to me] Vinyl sounds warmer and resonates more (feels more like listening to live music).
Either way, you get what you pay for. If you spend £500+ on a digital or analog system you're going to be in for some great experiences.
Not inclined more either way (they are both fantastic formats), just think we're lucky to have the choice.
Can I ask the writer?
Is there any audible difference between 16-bit and 24-bit? 44k and 96k-- to me there most woppingly certainly is! CD is limited to the latter; let us also remember CD is a rather old format. As for the viewer speaking of "The Nightfly": A terrific album, I suggest you listen to it in 24/96, it will blow you away. Oh, and pick up "Morph the Cat" while you're at it.
I grew up with CDs and listened to them extensively in my youth; yet I tend to find more richness in the sound delivered by records. To this day, I have not heard recorded percussion notes that sound anywhere near as sharp as they do on a 45 long play.
There's a reason for this; Turntables do something that DAC's can't. They reproduce exactly what's pressed into the medium.
CDs may have a greater resolution, but the sampled audio generally creates a notchy waveform. Therefore, most DAC's within a given CD player will apply a smoothing curve between sampled points. This curve is an ESTIMATION that's performed on the fly by hardware. This fact alone isn't something that's not particularly suited to discerning tastes. Yes, 44.1 kHz of 16 bit sample may seem like a very large number, but the smoothing + errors are clearly discernible in the overall recording. I have to say that I'm very happy that CD players come with a DAC that's capable of smoothing. Back in the early 80's, only the very top models offered this function. Smoothing certainly beats a notchy waveform.
If low SNR and maximum dynamic range are all that's important, stick with the CD. If you prefer richness and honesty, give records another try. Using a Pro-Ject Debut carbon with an Ortofon OM10 stylus, I find that the sound of records (of good quality and of a decent, cleanly condition) is crisper than anything digital. The audio is also more elastic and natural. It emulates guitars and other stringed instruments perfectly - especially the bass guitar. If I close my eyes, I almost feel like I'm at the concert.
Listening to digital music afterwards is like listening to muffled music. Sounds a lot smoother, but not nearly as sharp.
Part of the debate here should focus on content. All of the recordings that sound better to me on vinyl as compared to CD are from the era when the recordings were mixed to sound good on vinyl. This doesn't mean they couldn't sound better on CD, but to achieve that they would have to be remixed. The most obvious difference is the overall sound-stage.
With vinyl playback the channel separation is 30db maximum and that's only with a very small part of the audible spectrum. With digital it is of course much higher than that (90db+). So if you were mixing a master tape back in the vinyl days, to achieve a wide enough sound-stage you would have to do much more extreme panning. Now if you take that same master and transfer it to a CD, there will be big empty spaces in the mix that make it sound less like a live performance. For an example of how remixing a recording from that era can improve the overall sound-stage, pick up the CD of The Beatles Yellow Submarine Songtrack (not soundtrack).
The debate about ear fatigue is something different though. It would be difficult to get unbiased results, since people would know even with the best pressing and equipment that they were listening to vinyl. If the test was done with CD's that were burned from the vinyl they were being compared to then you might be able to get some conclusive results.
We could talk about what comes after the S/H -- the A/D converter that turns that voltage into some number of bits, but that's just a question of how much signal to noise ratio you require. If 144 dB is enough for you, let's go with 24 perfect bits. If you think you can hear more, we can add more bits.
But that's a separate issue from the sampling. Now that we've sampled our audio at the insanely high rate of 10 GHz, so that for all practical purposes it's a continuous analog signal, we have to ask ourselves if such a high sampling rate is really necessary.
Mr. Nyquist proved that no, it's not. In fact, we only need to sample at twice the highest frequency of our input signal in order to EXACTLY reproduce it. Exactly means both amplitude and phase.
To determine how fast we must sample, we only need to know the highest frequency of our input signal, and we must intentionally attenuate (by a very large amount) anything above that frequency, to avoid aliasing.
I will grant you that there may be frequencies well above 20 kHz that matter to us, even if we can't explicitly hear them. We can define the audio band any way you like -- 20 kHz, 40 kHz, 60 kHz or whatever. But once we do that, and our sampler meets the Nyquist criteria, we then know with certainty that we can precisely reproduce the original analog signal -- both in amplitude AND in phase.
I have to disagree here.
The sound being picked up by the mic, amplified and then sampled is, in your example, two waveforms of equal frequency but differing phase. As you mentioned, that phase difference is what provides spatial cue information.
But an ideal sampler (sample & hold) does not average anything -- it simply takes a snapshot of the voltage at its input at a precise instant in time, at each active edge of the sampling clock. That sampling clock can be as fast as you want it to be, within the limits of technology or your budget.
At this point, we haven't digitized anything yet -- we've simply captured the analog signal (voltage) at discrete points in time, and held it at a constant value until the next sampling time. If that -- holding the value constant -- is troublesome to you, then lets increase the sampling rate so those sampling times are closer together. The faster we sample, the more closely the voltage held on that capacitor in the S/H resembles an unsampled continuous analog waveform. If I still haven't convinced you this is true, then let's increase the sampling rate some more. Nevermind 192 kHz or whatever, let's go for 100 MHz, or maybe 10 GHz. Now do you believe the voltage on that capacitor is for all practical purposes indistinguishable from a continuous analog waveform? Remember, this is audio, so the high end of its frequency spectrum is far lower than 10 GHz -- much closer to 20-something kHz in fact.
Blog Doing Math in FPGAs Tom Burke 7 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...