About so - mp3 compression system first removes sounds that it believes are inaudible, and then as the compression rate is increased, proceeds to other sounds, creating the characteristic "chime" of dreadfully packed audio files. However, the fact is that even if the voice might be inaudible, it doesn't mean they don't contribute to the actual hearing experience. Infra- and ultrasound both do have effect on human brain, and that's one of the reasons why most people dislike organ music recordings. The actual organ concert includes so much more frequencies than any speaker system can play accurately, that the recording is often actually annoying instead of nice-sounding. Most other instruments sound just good in recordings, but organs actually use the whole church as part of the instrument, and the sheer size of the echo chamber creates some awe-inspiring stuff resonating from the walls.
That said, most cheap speakers can't play the "lost data" in mp3's anyway due to limitations in frequency area. But the better speakers, the more clear the difference between unpacked good quality wav or CD voice and mp3 made of it is.
Obviously, most people can't hear the difference between a good quality 320 kbps mp3 and 48 MHz 24-bit wav sound regardless of the equipment, but when the compression rate gets higher the difference is rather clear. 128 kbps mp3 tracks suck monkey ass compared to the CD tracks they are made of. The difference becomes most apparent in classical music records, pop/rock music doesn't suffer as much - mainly because classical music has usually so much wider frequency area.
Non-fixed compression rate formats are better than fixed-rate mp3's IMHO. They use adaptive compression that produces smaller quality loss and more optimized size.
