mgod wrote on Fri, 18 February 2011 16:01 |
Up is down.
Jitter was a solved problem, because it didn't exist. And certainly didn't affect the internet.
Then it did, and explained why things could sound better, and knowing that, was addressed. But it was solved before it was found.
But, what the hell: down is up.
All known knowns are known.
|
What are you talking about Dan?
Jitter is an obvious issue for anyone who looks at the mathematics (or even the basic process) of sampling, it's a given, samples have only two parameters, level and time (within the stream, not absolute time). It's a bit hard to not notice the importance of it, so if you think for one moment that nobody knew about it, you're very much mistaken.
However I don't know where you think the internet comes into it.
Unless you're talking about transmission jitter that is so great that it exceeds the ability of any buffering to absorb it, but once again that's an obvious issue (it's actually an issue with the streaming of any data, or indeed the streaming of anything at all, physical objects included), and it's also a rather different issue to that of jitter in the DAC.
Some things are known, some are basic common sense. Low level jitter is an issue in sampling, at two points, and two points only, the point at which the sample is taken, and the point at which the sample is used to reconstitute the signal. In between those the only way it matters is if it results in two systems being so out of sync that the data isn't there when it's needed, but electronically speaking that's quite a big error (you don't get a slightly wrong sample, or a sample at slightly the wrong time, you get a complete garbage sample or no sample at all), and the very fact that you're reading this tells us it's not happening over the net.
Think of it like this. Let's say you like milk on your cereal every morning, and so you have the milkman deliver milk (don't know if you get that in the States). So every morning you get up at 6 am, open your door, and there's a bottle of milk on your doorstep, you take it, pour it on your cereal and into your coffee and so you're happy.
Now, it doesn't matter whether that milk arrived on the doorstep at 5:59 or at 1 am, it doesn't matter if it was delivered to the milk depot at midnight or 6 pm the previous day, all that matters is that it's there when you open the door, every morning.
Now if one morning it's not there on time, this isn't a subtle thing, your cereal won't be slightly less milky, it'll be dry cereal, this is a catastrophic failure.
Now, let's extend it and say that rather than drinking the milk yourself, you're giving it to a baby, and this baby by some strange genetic quirk has a super accurate internal clock and likes its milk at exactly 6:30 am.... get it right and it'll be nice and quiet, the more you deviate from it, the louder it will cry.
So you check the time on your watch, and give the baby its milk every morning at 6:30 on the dot, according to your watch. Your watch however isn't perfect, it loses a bit one day, gains a bit another, on average it's right, but it drifts, it has jitter.
So the baby cries, this is where the degree of jitter matters beyond that of pass or fail, get a better watch and it will cry less.
But note that though it's crucial what time you feed the baby, it still makes no difference what time things happened further on up the line. The milkman could have arrived at the last moment or hours early, you see a package passing through a system has no memory of jitter happening in the various stages of its journey, whether it's a bottle of milk or a piece of electronic text or a sample.
To claim otherwise is to enter the realms of fantasy and metaphysics.