Slashdot Mirror
Vinyl To Signal the End for CDs?
PJ1216 writes to mention that vinyl seems poised to make a comeback in the music industry. Some are even predicting that this comeback coupled with the surge in digital music sales could possibly close the door on CDs. "Portability is no longer any reason to stick with CDs, and neither is audio quality. Although vinyl purists are ripe for parody, they're right about one thing: Records can sound better than CDs. Although CDs have a wider dynamic range, mastering houses are often encouraged to compress the audio on CDs to make it as loud as possible: It's the so-called loudness war. Since the audio on vinyl can't be compressed to such extremes, records generally offer a more nuanced sound. Another reason for vinyl's sonic superiority is that no matter how high a sampling rate is, it can never contain all of the data present in an analog groove, Nyquist's theorem to the contrary."
Too bad you'd need a US$10000 player to prevent your vinyl from wearing out. I for one would prefer properly mastered losslessly compressed audio files (or CDs if need be).
Years ago, when CDs first emerged I picked up a few Telarc disks and was impressed. Stupidly I assumed this meant all CDs would be of high quality and began physically downsizing my music collection. At some point, after unloading some treasures I'll never see again (for less than $$$$ on ebay anyway) I listened through a few recent exchanges and realised a lot of CD re-issues were shite. Bollox! I halted the exchange and have since retained the majority of my vinyl collection and even added to it. Some of that old well mastered stuff is well beyond the means of modestly priced CD player and even some immodestly priced ones.
A feeling of having made the same mistake before: Deja Foobar
Actually the statement about Nyquist's theorem is poppycock. This a mathematical fact, not some weird subjective result open to interpretation. Saying that Nyquist's theorem is wrong is equivalent to stating that the value of pi is really 6.
As you said, the comment about compression is nonsense. Compression is the removal of dynamic range, and is actually REQUIRED for vinyl to get the low volume sounds out of the vinyl surface noise to make them audible.
The truth of the matter is that vinyl records are crap compared to CD's in every measurable way - distortion, dynamic range, frequency response, signal to noise ratio, you name it. Are they perfect? No, that does not exist in technology. The Redbook standard is a tad short of the maximum theoretical dynamic range and frequency response the human ear is capable of. The conversion of digital data back to analog is tricky to get right. But it is superior to vinyl.
But some people do like vinyl better. Audio tastes are funny. People become habituated to certain types of distortion and other artifacts in the sound. To them is sounds better. But by any measurable means it looks like garbage compared to CD.
Given the speakers, headphones, and rooms that most people have, any argument about media quality is utterly pointless.
This is simply a media stunt by the recording industry's marketing departments to try and popularize a physical object that people must pay for. Vinyl has all the sex appeal to become that object.
Vinyl needs to be compressed even more so than CD's, as heavy bass can be enough to make the needle pop right out of the groove.
However, all the arguments about increased sound quality, as you point out, are absurd.
I am a mastering engineer, software engineer and have worked on audio software. And in all of my experience there are only a couple of things left to improve upon with current digital audio technology, but for a very small amount of return.
When the music is mixed digitally using certain "professional tools" (no pun intended) it is done in fixed point. A few companies have realized that using double precision floating point *does* sound better. And the difference is measurable. Some sound engineers believe it's also very audible.
In short, sampling a signal, scaling it, summing it and then truncating (or dithering) it, does more than shifting it's level and burying the lower end under the quantization threshold. No technical name exists for this type of distortion, but it is a self correlated noise upon the signal, or cross correlated with the other signals being mixed upon it. What it amounts to is to putting the signal thru a transfer function consisting of a jagged diagonal line (instead of a perfect diagonal line, whose slope matches the gain applied) or jagged grid that shifts up and down with the value of the other streams being mixed. This is analogous to rendering a diagonal line on a computer. The higher the resolution (number of bits) the better. But sadly, at the recording and mixing stage, mixing a large number of tracks with say 24 bits of fixed point resolution is ridiculously bad, even if the final master will be dithered and truncated at 16 bits, because this distorting process will occur repeatedly, for each gainstage, for each track summed. One solution to this is to apply gain and sum at double precision floating point. Yet another, less popular solution, is to actually reproduce each track back into the analog world using high quality DACS and sum in the analog domain. Both sound nearly as good, and certainly better than summing at 24 bits fixed point.
Second, there are certain IIR filters that can't be implemented at just 2x the bandwith. Because of this, the choices are: Upsample and downsample just for that filter (which is computationally expensive and if done at all, seldom done correctly) or just run the entire audio stream at 4 or 8x the bandwith.
What is done today by most studios is run the entire project at 88.2 or 96 kHz sampling frequency. This is great, but requires a very high quality downsampler at the end of the chain to convey the final result.
One could argue that vinyl masters can be cut from a DAC running at 96 kHz and thus have an increased frequency resolution. But that improvement pales in the light of the background hiss level, additional bass compression required for vinyl, preamp distortion, de-emphasis equalizer tolerances, motor speed stability deviations, etc.
I wonder if we just had a tiny speaker on top of a CD player reproducing the very high frequencies that come from the "needle" whether it would finally pass for vinyl.
I bet that much of what is perceived as sounding better for vinyl is the fact that people can hear the sound of the mechanics (the needle itself) as well as the speakers. I remember as a child, that the records sounded a lot better when the turntable lid was open.
-arr