What may be good for a voice podcast, is not good for music.

Ever listened to a sermon podcast in the car and got frustrated fidling with the volume control when your pastor goes from really quiet to really loud or visa versa? To deal with this problem we post-process our sermon recordings with a software compressor, effectively making the recording all one level. (Actually doing this with software [we use audacity's compressor effect] as opposed to hardware at recording time introduces it's own complications and sometimes we still don't get it right. Primarily this is because it requires investigating each recording and then applying the appropriate compression ratio's.)

However, making a recording all one volume pits listening convenience against dynamic quality.