Mixing for mp3 output

[john caldwell]

Only recently have I made attempt to mix toward an mp3 output. Mp3 files have been useful for the obvious reasons that small files offer, but I've been suprized by some of the changes in mix balance when bouncing to disk using the Pro Tools mp3 codek, typically at 192 KB/sec. Critical midrange elements appear to receed into the mix of mp3 renditions, as compared to uncompressed bounces, for example.

What generalizations, if any, can be made about the mp3 compression process? Further, can anyone volunteer changes they deliberately make in their mix/master workflow when anticipating an mp3 output - possibly analogous to mixing and mastering for FM radio and the way that might call for a different approach than mixing for CD?

Thanks group.

John-

[Lawrence]

I don't know anything about the codec in particular but experience tells me that a good mix should survive a 192k mp3 translation pretty much intact and with very close to CD audio sound.

Have you tried using another audio program to render the mp3 to see if it makes a difference? Have you tried rendering the mp3 directly from the multitrack mix inside PT instead of printing a mix and then making an mp3 out of that mix file? Or is that how you're doing it now?

When I finish a mix in SX I render a 24-bit master and an mp3 directly from the multitack mix "first generation" (don't ask...). It seems to sound better to me than loading the 24-bit master back into the daw afterwards (or into Wavelab) and rendering the mp3 from the mix file.

Honestly... I make mp3's at 192k only and they always are a very good representation of the mix, for better or worse. 128k sounds a lot worse to my ears.

Lawrence