Yesterday we summed up how PCM encoding works in the broadest of senses. I used the moving train analogy hopefully to good effect but my friend Len Schneider sent me yet another one that I think can only help our understanding of this subject. It's worth getting another summation.
"If you slice a loaf of bread, weigh each slice on a very accurate scale, and then note each slice's weight on paper, you've "digitized" that loaf of bread.
When you slice, you're "sampling" the whole loaf. When you weigh each slice, you're "quantizing" each slice by assigning it a numerical value. When you're done, you can add up the weights of each slice and determine how much the whole loaf used to weigh.
Of course, you'll end up with some inaccuracies. If your loaf is too big to fit into the slicer, you'll need to cut off a few inches. In electronic terms, the equivalent process would involve filtering out high-frequency content that the digital system can't handle—and that's exactly what an A/D converter's "anti-alias" filter does.
If the blades in your slicer aren't spaced exactly the same distance apart, you'll end up with uneven slices. (That's a crude description of "jitter," or timing inaccuracies in the initial sampling process.)
If the blades aren't sharp enough, you'll probably end up with some “crumbs†(conversion artifacts we really don't want).
If your scale isn't accurate enough, you won't be able to assign a precise value to each slice. (That's called "quantization error.")
And if you can't weigh each slice accurately, you'll never be able to refigure the exact weight of the loaf before you sliced it in the first place."
I think that's a great analogy. Hopefully it'll further shed light on what's going on.
I want to go back to my train analogy for one last thought this morning: jitter.
Remember in the moving train analogy I suggested that if you took a series of snapshots of the moving train you could examine each picture as a fixed point in time - which is what happens with a sample. But I also mentioned that if you played back the series of snapshots quickly enough you would recreate the moving train again and it would look like it was seamlessly moving - which is how a motion picture works.
But if you've ever been to a movie theater and had the film skip, the illusion of motion is instantly broken and you get a distorted motion picture. That is, in essence, jitter.
Jitter happens when the sequence of samples gets out of time. The illusion of motion in a movie is dependent on each frame of the movie being projected at exactly the right interval relative to theprecedingframe. The same is true for music and audio. The samples all have to arrive in perfect timing to each other. A slow moving change away from this perfect timing is hard, if not impossible, for the ear to hear. But quick changes back and forth in time are easily discernible by the amazing ear/brain relationship.
