When considering different
compression algorithms, for either
text,
images or
audio, the compression ratio is the ratio of final file size to the original file size.
For example, if I have a text file thats 1000K, then run bzip2 on it (probably the best widely available text compressor), it's 200K, then the compression ratio is 5:1.
Some compression algorithms are lossy, typically multimedia ones, e.g. JPEG and MPEG. This means that when things are decompressed, they are not exactly the same as when they were compressed. This usually allows a higher compression ratio. Some algorithms allow quality settings to trade off quality against compression ratio.
In general, the compression ratio depends to some extent on the input. Some things, such as, say a very repetitive text document, will compress very well; while others, say a phonebook, which doesn't have that much repetition will not compress so well. So, for example, bzip2's typical compression ratio is anywhere from 2:1 to 5:1, depending on what you try to compress.
However, a recent trend for some applications is fixed compression ratios; automatically fixing the compression ratio, then trading off quality as necessary. For example, MP3 (more accurately: MPEG-1 Audio Layer III) audio encoders usually have a fixed compression ratio. If you compress a stereo 16-bit 44.1KHz (i.e. CD-quality) recording to 128kbps MP3, it will be an 11:1 compression ratio. MP3 does have support for variable bit-rate technology (i.e. for bits that need lots of aural detail, you bump up the bitrate ... i.e reduce the compression ratio), but nobody seems to use it.