Law 15 applies
There are several different kinds of
compression.
Compression in the analog world (where the DIFFERENCE) between the loudest and softest passages are reduced (separate parts of the frequency domain can have different levels of compression).
This type is/was used to fit the dynamic range of a studio recording onto vinyl (which has a comparable resolution of around 11 to 12 bits) and used in dBx, Dolby (A, B, C, S & HX-Pro)
Analog compression is still used in some car audio, FM, AM radio, vinyl releases.
Also non linear tube amps exhibit a (very slight) form of compression.
This affects the sound.
In small quantities it can enhance the sound and most people find compressed audio MORE dynamic... why because they can hear the nuances (which ar low level) more clearly because they have become louder but not too obvious.
Nowadays compression isn't done with analog techniques (amps with a gain that varies depending on the audio level) but completelty done in DSP where it can be done in a better and less obvious way (upto a certain point).
You can take this MUCH too far as the loudness wars point out.
Many non audiophiles do not even hear this and think it sounds nice !
So pure analog compression and a digital (DSP) version of it.
Digital compression is another thing and if lossless the INFO in the file is reduced by clever techniques that allow bit patterns to be substituted to smaller sized patterns that the decoder recognizes and upon decompression simply replaces with the according (original) amount of bits.
So a smaller amount of bits with the acat same 'info' but decsribed in different way with less bits (and thus less storage room needed).
The analog represented signal is not changed nor compressed like analog compressors do.
Different formats compete (by clever technology) for the smallest sized files.
How small a file can become depends on algorithm and dataformat that can be made smaller.
a jpg and FLAC are already compressed and won't benefit from compressors like ZIP and RAR but WAV and Word files e.t.c. can.
different Audio formats (lossless) exist to avoid brands having to pay for the usage of formats so they make their own.
Audio formats are can compress more than data compressors eventhough they work the same.
This is simply because it is designed to cope with audio datawords and can be used 'on the fly' and can include certain other data as well (pictures, lyrics, additional data).
They are not there becasue they sound better but to prevent having to pay royalties.
lossy compression is done by analysing (very short) pieces of audio (described waveforms) and decisions are made (depending on quality and desired file size) what may be discarded as relevant info.
This also is NOT dynamic range processing as in analog.
This process results in a less complex waveform description that can thus be described with fewer bits (so the number of bits, not bytes) is reduced.
Afterwards digital compression (as in lossless) is also applied to further reduce file size.
On decoding the simplified waveform is retreived again (and before it goes to the DAC) is PCM again, but might be missing harmonics or tiny details at certain passages that are considered to be masked by loud passages or are not really contributing to the essence of the music.
So compression is a dangerous word that causes LOTS of confusion and can describe many different processes/effects.
CD's are not compressed but expanded before it is pressed onto a disc.
So a LOT more bits are on a CD than there are bits in the audio that is represented.
This is for redundancy, tracking reasons, added info (number, timecodes e.t.c.) and limitations of the CD format.
It involves padding (adding extra bits for certain reasons) and error correction (CRC, added bits that can be used to calculate what bits were intended to be there upto a certain point of 'lost bits') and spreading chunks of bytes over a larger surface (scrambling) so when a disc is damaged the original bits can be reliably retreived (accurately calculated not quessed).
Also extra bit's (even bytes) are added for synchronising purposes.
Jitter has no influence on the DATA itself untill it is SOOOO extreme bit values are altered (0 becomes 1 and a 1 becomes 0)
The 'sounds different' debate (that is not recognised by any technical experienced but ignorant and closed minded people that have no idea what Frankenstein they actually created) is the suspicion that the described data word is NOT the only thing that determines the reproduced analog waveform in the end but that maybe the actual (digital) bit waveform itself or it's timing also influences the analog waveform in a way that can (at present) not be seen or measured or analysed but only be heard.
As far as the timing is concerned there is some influence on the actual reproduced waveform in the DAC circuit (cheap or older DAC's without jitter reduction or not well performed jitter reduction).
The analog output current/voltage may (bad implementation of current sorce regulation or PCB layout) may also cause noises to be 'added' on top of the signal.
At least that's what I can make of it but am rather limited in my way of thinking (being closed minded and all).
Problem is for it to be accepted by the computerworld some evidence must be presented in a scientific way so the one track minded engineers that cannot see past their 'ideal' digital numbers world can understand WHAT to look for.
Unfortunately this requires supplying proof of some sort, by a tech controlled test.
Only verification, even if by Obama himself, isn't going to cut it.
This is never going to happen though for various reasons.
So IF you can hear it... use it and enjoy it.
If you find a reason... present it to AES or Hydrogen and show them they were wrong after all.
Like a teacher used to say (when he didn't know how something needed to be explained):
Vraag me niet hoe het kan... maar profiteer ervan.(Don't ask me how it is possible, but enjoy the benefits)
It will stay in this category for ever I am afraid.
But it's a typical remark of an ignorant,biased and closed minded EE with bad hearing and not resolving mediocre quality equipment.
So be it.