If you encode at a reasonable bitrate with a proper encoder(this is key), you'll never notice.

I would, however, use a better codec. MP3 is over 20 years old. Use Vorbis, or Musepack; anything performs better than MP3. A lot of modern players are starting to support these codecs. My Rio Karma supported Vorbis and FLAC since 2004. They do much better on blind listening tests than MP3 at lower bitrates, and at typical bitrates do much better on test clips that MP3 fails miserably on.

It wasn't until I modified a PCM1716 based D/A converter with a good output stage & linear power supply that I could really tell a difference between certain MP3s and the original file. Even after I did this, a good LAME --preset standard MP3 vs. the original WAV was something I could not determine with better than 12/16 in a blind test, and neither could a lot of the people listening. This is on a good system in a decently treated room, so I wouldn't go nuts about it.

With the D/A in this yamaha receiver I had before, none of that mattered. It was like a veil. This current D/A had a 4558 IC for the output stage, which is like a distortion pedal(probably why it's so often used in distortion pedals!), and masked lots of artifacts present in 128k material. The new DAC with new output stage, new film coupling caps from PCM1716 to output IC, bypass caps on output stage, and new PSU cost $115! When I saw a PCM1716 based DAC going for $79.99, I knew it'd be worth the time to open it up. It beats a lot of the audiophile esoterica nonsense I've seen going for $1500-$5k & up in standalone CD players that use 5 year old D/A chips and output stages with 5532s from 1976.



Last edited by Thasp; 04/30/09 04:12 AM.