What does "compression" like?

I often hear the term "compression" used as a negative in audiophile-oriented music reviews, but I don't have a good handle on what it is or what it sounds like.
Enlightenment, please?
compression is necessary to deliver quiet parts with more volume and loud parts with lower volume accordingly.
the recording always have issues with either low level that is comparable with the noise floor or upper level that goes beyond loudspeaker dynamic range.
Hi Reb.

Compression usually refers to compression of dynamics, where the softest sounds are louder and loudest lower than "normal".

How applied or not is part of how most any recording is made.

Its usually done to make all parts of the recording able to be heard more consistently, especially at lower volumes and/or when external background noise is present..

The negative for audiophiles often comes in as reducing the "jump factor" often associated with good dynamics, and also distorting the presentation of individual acoustic instruments compared to live playing.

The big negative that can come into play with dynamic range compression is when waveform peaks are clipped off as part of the process. This is the most significant and generally offensive kind of distortion often but not always introduced as part of dynamic compression.
Agree with the above posts. A good example of what compression sounds like would be if you listen to a classical FM station on your tuner for a big orchestral piece like a Mahler symphony. At the climaxes where the whole orchestra is playing full blast, you'll notice how the volume drops dramatically. For a contemporary pop recording, as a general rule that won't happen because the recording itself has had its dynamic range compressed, so the difference between loud and soft parts is minimal.
If it always sounds loud, no matter the volume of playback, it is compressed.
Compression comes in two forms. Digital file compression, such as MP3s. It discards information that for portable, uncritical listening at low levels won't be disturbing. However, today, when you can hold gigabits of data on a USB drive the size of a fingernail, there's no longer any need for file compression.

The other compression is dynamic compression. This also is no longer necessary for most digital recordings. 24bit provides a 144 dB dynamic rangeĀ—that goes from just barely audible to painfully loud and hearing damage. There's no reason for music recordings to be compressed, because, for example, the range of sounds you'll hear from a symphony orchestra, from the softest solo instrument to the loudest passages with all the brass blaring and the percussion going full force, spans about 60 dB. Compression kills the expressive dynamics that musicians painstakingly put into their music. It takes away the transient spikes that give percussion its punch.

How it sounds is difficult to describe, because it doesn't produce any gross distortion that would make you cringe.

Here's a link that has both a visual and audio example of dynamic compression, plus a couple of additional links.
Made an error in the url, try this dynamic compression.
This is not an answer to your question, Rebbi, rather just a side-note. We all hate the compression used to affect dynamics for ipod and radio play.

Compression is also used in the recording studio and not in the negative ways that have been mentioned. A compressor would typically be used in recording Rock music to shape the sound of an instrument. For example, by applying compression to a bass drum or tom tom, you can affect the decay of the sound; if it is too boomy or has a long decay, adding a little compression can change the sound of the drum.

This type of compression along with other effects have been used creatively in the studio for many years and is a tool used by the producer and engineer. Usually you would never even know that it was being used.
Making recordings is an art not just nuts and bolts engineering.

Compression is one of the tools applied as part of the art and not always necessarily a bad thing. It's a tool that provides various views into the music as envisioned by recording engineers. Sometimes, that vision is a purist one, attempting to reproduce what was heard live, but that is fairly rare these days but still practiced well by those with skill who choose that style.
Data compression algorithms for digital files are entirely different than compressor/limiters techniques used in audio processing. For nearly all pop/rock recording styles a little compression makes the recording sound better.
Simply, if you notice on different recordings, the volume knob for one recording, you barely need to turn it on and the music is very loud. Other recordings you can turn the volume knob half way before it gets loud. The one that you barely have to turn the volume on is what we would call very compressed. Especially if you are familiar with a song live, if you know the beginning is soft but on a recording it is loud, then the volume has been raised for the beginning. Basically the whole song is at one volume level and any changes in volume during the song (dynamics) are gone.
Like Wildoats said. Plus if the singer is the same volume as the drums and guitar.

I've seen a few user reviews on Amazon where the reviewer says that something's wrong with the CD since he needs to keep changing the volume during the song.
I always feel like making a comment, but it might get ugly.
Thanks a lot for all the educational replies, folks! Now I understand. And what I
was asking about was dynamic range compression, not file size
("lossy") compression of digital files.

Again, very helpful!

Just as an aside, not all file compression technology is lossy.

FOr example, FLAC files can be compressed without losing any digital data, although compressed file sizes will be larger than lossy compression, like MP3, but smaller than uncompressed.

I just converted my music library from uncompressed .wav files to lossless compressed FLAC. Overall size went from ~ 900+ Gb to about 500Gb, which, based on my personal professional experience in the past developing computer image data compression algorithms, I would expect to be typical lossless compression ratios for most digital media content.
Two threads in one here.
Mapman: right, hence, for example, "Apple Lossless" codec.