Here is my incredibly simplified and clumsy explanation. It’s not technically correct, but I believe it gets the point across. We know digital music is stored as bits. These bits are 1’s or 0’s. Sometimes the 0’s are silent parts of the song and sometimes just number fillers like the zeroes in the number 100. But in any event, there are lots of them in a music file. When we save the file in a lossless format, every one of those zeros is also stored. Most of us have been to a sporting event, so the following picture should look familiar.
Your eye should see the number four. If we saved this digitally, we might use 1’s for the black dots and 0’s for the white dots. If we agree to always go from top to bottom and left to right, we would get a file that looked something like this:
10010
10010
10010
11110
00010
00010
00010
By using this file, it would enable us to perfectly reconstruct the number four on our ‘scoreboard’ by turning on a light every time we see a 1 and skipping a light every time we see a 0. It takes 35 bits, because that’s how many we started with.
Now let’s say we substitute the number of consecutive 0’s we run across with that number, so we have a file some thing like this:
1210
1210
1210
1110
310
310
310
We would turn on the first light, skip 2, turn on a light, skip one, etc. This would still enable us to perfectly reconstruct the number four and only take 25 bits. We saved 10 bits out of 35, or 28% of our file size. (I realize that the numbers 2 and 3 aren’t 0’s and 1’s, but I told you it was simplified). It should be easy to imagine that we can save more space when ‘encoding’ the number 1, than we could save when ‘encoding’ the number 8. The same is true when we encode a song with a lot of quiet parts as opposed to one that’s non-stop music. That’s why one 4 minute song on your album might turn into a flac file that’s 652kbps, and another 4 minute song is 1005kbps.
Go look at the picture again. If we chop off the entire top row of ‘music’ we can still ‘see’ the number 4. That is lossy 'encoding’. Something is missing, but we can still tell it’s a 4. We might think the scoreboard was made that way if we didn’t know better. Not that big a deal. Imagine 320kbps mp3. Now picture the second light in row 4 goes out on our ‘scoreboard’. Much more noticeable. We can still tell it’s a 4, but we can tell a light is out. Picture 192kbps. A few more lights go out and we start to have a hard time reading the score (pun). Picture 128kbps.
Now in all fairness, the guys who make the lossy encoders are pretty smart. They know which parts to leave out that are harder for us to detect. Picture you are talking to a buddy in the parking lot and an airplane goes overhead. He doesn’t stop talking, but you don’t hear a few words clearly because the plane drowns him out. Lossy encoding throws his words away. The loud noise ‘masked’ the quiet noise. Many people don’t notice. Depending on what your friend was saying and how closely you were listening, you may notice. In today’s world, storage is so cheap and internet speeds are so fast that there is really no reason to go lossy. I just bought a 2TB drive for my music library to grow into for $80. I save my music in flac file format because it saves some space but doesn’t sacrifice any quality.
I realize this explanation was pretty hokey, but maybe it’ll help somebody to get a basic understanding or motivate them to go read up on the real stuff.
