Be a little more scientific about the question.Rip the CDA file off the CD and convert it to .wav format.Import both it and the original recorded wave into sonor and zoom almost to the sample level to compare for any artifacts of the CD burning process.Use both monitors and phones to A-B the two tracks.Are they really different?If so something is wrong because digital transfers of information are supposed to be pretty darn exact.
Compression is commonly used with an outboard unit when tracking any sound source where there is a considerable variation in volume.Look at the waveform of a recorded sound and you will see peaks and an average level.The difference between them,the peak to average ratio,is a good starting # for the threshold setting of the comp.Compressor Ratios of 2:1 to 4:1 or so are pretty mild.Anything over 10 or 12 and you are into limiting territory.But as opposed to trying to dial it in by the numbers,use the gain reduction meter so that between 3 and 10 dB is the average.You really want to just compress the peaks.Once you start dipping down into the body of the program material,you will hear pumping and breathing artifacts.
After the tracks have all been recorded,you may wish to use software plugins for further compression to smooth out the sound.Since I switched from 16 to 24 bit recordings,I don't compress the front end any more.The extra headroom makes that unnecessary and gives you dynamic range (hint).I do continue to compress in software,I just don't need the safety net of a compressor with 24 bit to avoid hitting zero dB.