I've spent a semester in a high-school recording studio (We produce our own compilation CDs each semester recording bands formed by kids in the school), and have done quite a bit of research on both analog and digital audio, various formats, and codecs. I don't claim to know everything, but I think I know enough that I could help some of you guys out.
If there's anything I don't know, then I'll also be willing to search for the answer (Though I can't guarantee results.)
I'd like to start off by explaining sample rates.
When we talk about audio, we mention 2 numbers most commonly, the bit rate and the sample rate.
I'll assume you know what a bit rate is, because it's a pretty simplistic concept.
On to the second number. The term sample rate may not make sense at first, and you may be wondering what exactly a "sample" is. I'll have to explain a little bit about how computers store sound to explain this. Don't worry though, I'll keep things as simple as possible.
When a computer records music, it can't just store the sound the way it came out of the instrument and into the microphone. That's analog sound. A computer can't use analog sound because everything inside it is done digitally, using whole bunches of 1's and 0's to describe everything. Now, without getting to far into specifics, I'll explain where the term sample rate is important.
As a computer records the sound, and turns it into digital 1's and 0's, it can't just record it as a continuous sound, even though it is. 1's and 0's could never describe every single minute detail the way our brain can, so the computer has to break the sound down. The way it does this is by making a whole ton of "samples". Each one of these samples essentially tells the computer where the speaker cone should be. How far in or out the speaker has to move. Here's a small diagram for you:
Each small break in the lines there would be where the computer takes a sample. That line illustrates sound vibrations in the air, and the blanks are where the computer would record the sound. Each of those blanks would be a sample. The speed that the computer takes these snapshots of sound is known as the sample rate.
Now here's where things get a little bit more technical. When you record something at a certain sample rate, some sound can't be heard by the computer. If the frequency of the sound is faster than half the sample rate, then the computer can't pick it up. This number, Half the Sample Rate is also known as the Nyquist Frequency. This is the highest frequency you can record at that sample rate.
For example, if you are recording at 44.1khz, the Nyquist Frequency is 22.05khz, which happens to be slightly over the highest frequency humans can hear (Which is about 20khz, and gets lower as you get older, or get hearing damage.) Regular music CDs are encoded at 44.1khz because of this. However, most of the time the music itself is recorded at higher sample rates. Music encoded at anything above 44.1khz is either a file made from a master copy at the studio, or has been "up-sampled" from 44.1khz. Since leaking master files generally never happens, and those master files also happen to be individual instrument recordings, any mp3 you see running at anything over 44.1khz is likely upsampled. Unless it's a home recording.
So, to quickly highlight some of this information:
- Music is recoded with snapshots called "samples"
The frequency that the computer records them at is the sample rate
The highest frequency recordable at a given sample rate is exactly half the sample rate (ex. 44.1khz/2 = 22.05khz)
The human ear cannot hear higher than 20khz
Music is commonly recorded at higher sample rates than 44.1khz, however, this is simply to have the highest quality recording to work with