Look at the music I like!
I like to listen to music!
I have been collecting music on vinyl and later CD and recently FLAC files since the mid '70s. Slowly I have built up a replay system I am happy with, for vinyl, CDs and even cassettes. I can state categorically that a properly recorded analogue vinyl record sounds way better than CD. The reason this is not more widely recognised is the vast variation in sound quality achieved by different qualities of replay devices. Whereas it is not automatically true that the more expensive your set-up the better it sounds, it is true that it will cost several thousand pounds in today’s money to buy a deck/cartridge/preamp source that will outperform the best CD players. But once you get to that level the difference between LPs and CDs is like night and day. At the cheaper end of the market a £100 CD player will sound much better than a £100 record deck, but that does not mean the medium is inferior, just that the replay device is limited. To all those who think CDs sound better, please listen to a high end record deck with a moving coil cartridge feeding into a top level phono stage, then connect that to a transparent valve based amp and clear speakers. then you might know what you are talking about. And make sure you are playing true analogue vinyl recorded before digital mastering and cutting devices became ubiquitous. Modern vinyl is usually a digital signal at source and is digitally processed during record cutting, and so can not demonstrate the mediums true power.
OK. I am going to put on my engineer hat (well, I never actually took it off). According to theory, if you put a signal into an ADC (analog-to-digital converter), you can take the digitized signal and put it into a DAC (digital-to-analog converter), filter the output, and get EXACTLY the same signal, plus a tiny bit of quantization noise. However faster sampling and more resolution can reduce the noise to practically nothing. (To simplify, I am going to ignore Nyquist here and just assume that we are sampling fast enough).
This means that, in theory, one could use a computer to digitally record the output of your record player, and play back that exact sound at any time from a FLAC file (I am assuming that you use pro-quality gear that can record/play at 24-bits & 96KHz). If you make no mistakes, theory tells me this is true.
That means that the simply act of putting music on a record and getting it off later somehow changes the sound. Now, if somebody were to figure out a mathematical model of this vinyl process, it should be quite possible to have a plug-in that can apply the vinyl sound to ANY music, and skip the vinyl entirely. You should be able to get the vinyl sound from any cheap CD player (plus the tiny bit of quantization noise that CDs introduce, but that should be imperceptible). Or, distribute a 24/96K FLAC file.
This whole thing is engineering and physics. No mystical forces in a record player.
Comments?
The Foobar2000 Convolver plugin would do that, if you had the right impulse file. I know that people did create impulse files using old valve amps.
I am bemused to note that my non-professional software (Nero Wave Editor)
has a “re-analogue” effect, with filters such as hiss, click, crackle,
buzz, and settings for type of record (33/45/78), and sliders for level
(and a few other settings).
As I said, non-pro SW, so I don’t know how convincing the effect would be
to an audio engineer, but it seemed quite good to me the one time I recall
trying it out.
I am not sure that convolution is the right answer here – convolution is completely linear. I would expect that some of the “good stuff” in the vinyl process is non-linear.
Still, it might be possible to get “close enough.” I am just not sure how you could get the impulse response of record player, though. That is beyond my capability.
That’s why the engineers made the spec for CDs what they did, but their measurements of master-to-vinyl were a measurement of an analog-to-analog process. Worse, it was a measurement done at the beginning of a product cycle back in the 1980s, it wasn’t some kind of truth-seeking scientific mission. Writing the “digital equivalent” as a particular number of bits at a certain rate makes engineers feel good, but it doesn’t necessarily match reality. Especially not in a product development process, where the goal is to put out a widget that can be marketed as “better” than vinyl while still being possible to put on the market at a large scale. Remember, sixteen bits was the state of the art of the time. No surprise that their product was magically “just a bit better” than the old technology.
FLAC’s maximum resolution and sampling is irrelevant. When it’s pulled from a CD it’s limited to CD resolution. FLAC is mentioned in this discussion because it’s the lossless format used to rip CDs.
You can, however, use FLAC to record a needle-drop, or if you like SACDs, a laserdrop. Naturally, the quality of your A/D converter matters.
When I want to listen to music turned up loud on my home system records sound great. most CDs make me want to turn it down. Noise is not the issue.
Since when is analog equivalent to a digital sample? It’s fundamentally different.
I don’t know how I can show what the resolution of vinyl is (I can believe that you are correct about the engineers getting it wrong in the 80’s), but my experience has been that it doesn’t seem to go higher than 32KhZ. It might not be noticable to me now but I could tell 12 years ago when I was transfering some music that I couldn’t get on CD. I certainly don’t believe that the resolution of vinyl is infinite, as I have seen claimed on certain audiophile sites.
It is true that you can only get out what you put in, but we are getting past the CD resolution now. Pono are claiming 24/192 for their FLACs, I am interested to see if they will deliver (Some of the audiophile comments around it have sounded like woo and have been a bit discouraging.)
I am not talking about mystical forces, or audio theory, just what my ears tell me when I listen to music. I have also recorded from LPs to 16 bit 96Khz, (as high as my soundcard will go), and on replay A/B comparison shows the original still sounds better. Its not a theoretical argument, I can hear that original analogue vinyl played on my set up sounds better than digital. If you could get to my pad in London, you could hear the difference too, its not small. And to answer your post, no a digital, analogueifying plug-in will not even come close to true analogue sound, although it can obviously add typical vinyl type artefacts like scratches and clicks. I think you have simply not listened to the best that vinyl can produce on sufficiently high end gear, with open mind and ears. Anyway enough of this, I am off to listen to the some original 70’s vinyl. Its the music that tells the story not theory or mystical forces.
You have that right. Personally, I can’t tell much of a difference between CDs and vinyl, and I don’t know if what I’ve noticed is because the analog nature of it is better or if it’s just smoothing some rough edges. I used to think that 1024 vertical pixels was all that was needed in a screen, but I think there’s some practical value to going a good deal beyond the “average” human specification.
This whole thing is just one cat and girl comic, isn’t it:
and
I could go on, but I’d post the entire run…
Edited to add another: http://catandgirl.com/?p=283
Know how a car owner can have no mechanical skills, but still recognize the sound of their engine changing when it’s wrong?
Digital tracks don’t change or personalize over time. Analog does, and people who love analog are being nostalgic, but that doesn’t mean they’re wrong. Those scratches and pops tell a story just like the music does.
Here’s an example:
When I was a kid, my brother got a copy of The Wall as soon as it was released to vinyl. We both played the album over, and over, and over. We did it until a scratch developed at “Comfortably Numb”. Then, whenever you played the album, you’d hear, “My hands felt just like two balloo, two balloo, two balloo…” until you lifted the needle and reset it.
It was entrancing.
If you were tired, it was hard to get up to fix, because the effect was so perfectly placed in the album and honestly not at all hard to leave as a background loop. My brother didn’t bother to stop playing it, and neither did I. He didn’t go get another copy of the album. His own worn and scratched copy was exactly what he wanted. That distortion of sound spoke of just how many times we’d played the album, and how much we loved it. Like a teddy bear missing an eye, or a leather jacket covered in patches - the vinyl spoke about more than just the songs on it. That’s what vinyl does that digital can’t do - it acquires history and personalization though use.
Well, since you asked (and you did), I am going take you to school. There are TWO important numbers in sampling: how fast, and how much. Sample rate is how fast, and the sample depth (number of bits used) is how much. They are separate discussions.
SAMPLE RATE
This is simply how many samples you take per second. According to the theory, you need to sample at least twice as fast as your highest frequency. If your music has sound up to 20KHz, you need to sample at least 40 KHz. Now for the assumptions. First, you have NO frequency content in the signal above 20 KHz – not a bit. Second, when you play back the recorded signal, you need to block ALL output frequencies above 20KHz. If you do this, the output will be EXACTLY the same as the input. The devil is in the details. The big thing here is that there is no such thing as a perfect filter – either on the input side or on the output side. There is no such thing as a filter that can pass signals at 20,000Hz and block all signals at 20,001Hz. Not possible. To get around this, you need some safety margin. That is why CDs sample at around 44KHz – about 10% over that Nyquist says you need. This is to give the filters some wiggle room. You can start to block at 20KHz, by the time the frequency gets to 22KHz, the filter has had time to kick in.
Now, if you have music up to 20KHz, and you sample at 96KHz, that gives you a LOT of wiggle room for the filters. It would be hard to design a filter that could NOT eliminate all frequencies above 48KHz if it could start at 20KHz. This means that your filters can do a far better job and you do not have to worry at all about aliasing (what happens if you break Nyquist’s rule)
Now, if you look at a reconstructed signal WITHOUT the filter, you will see a “stair-step” pattern. Check out the “DAC Output” graph here:
http://www.ni.com/white-paper/5535/en/
What you are ACTUALLY seeing is the original signal with some high-frequency harmonics. If you use a good enough filter, you can almost completely eliminate those “stair step.” With perfect filters, the output matches exactly the input.
Sample Depth
An ADC and a DAC have a finite number of bits. When you sample a signal, you grab the input at a particular moment in time. When you grab, you have to turn it into a number. How many numbers do you need? CD standardized on 16 bits. That means that you get a number between 0 and 65,535. If the actual voltage corresponds to 200.2356, it gets rounded to 200, giving you an error of 0.12%. The lost 0.2356 is the quantization error. If you use more bits, you throw away less. 24 bits (current standard for ultimate quality in the studio) uses numbers between 0 and 16,777,215. Scaling our input up to cover the new range, our new input is 51,260.3136. We throw away the .3136 so the error is 0.0006%. Each extra bit added will cut your error in half. Since we added 8 bits, our new error is 1/256 what our old error is. If you get your error low enough, it will be so small that you cannot hear it.
PUTTING THEM TOGETHER
So, when you sample, you can get errors from sampling (quantization errors) and also errors from inadequate filters. The more bits, the less quantization error you get. The higher you sample, the easier it is to get the filters to do a good job.
So, in theory, if you sample fast enough, and with a good enough filter, and use enough bits, the samples signal should be indistinguishable from the original.
One more reply…
Edited to add: As I see it, the repopularization of vinyl has little to do with fidelity, engineering, and what form of recording sounds best, and it has everything to do with the nature of human nostalgia. Plus, it’s driven by people in the industry itself, who are looking for a way to keep themselves in house and home, etc.
A somewhat clearer explanation than many … thanks.
Related point - if it’s not more effort than seems reasonable, what’s your take (or anyone else’s, for that matter) on “…no point to distributing music in 24-bit/192kHz format. Its playback fidelity is slightly inferior to 16/44.1 or 16/48,…” as seen at http://xiph.org/~xiphmont/demo/neil-young.html and elsewhere?
I can grasp most of his arguments, but lack the technical background and experience to reliably assess his conclusions. Thanks for any thoughts …
