If you happen to listen to your mp3's on your surround sound system you might want to try a few experiments, as I did, before choosing a baseline bitrate setting.
I found that music ripped at a constant bitrate of 160 or even 192 kbps could still produce strange "watery" sounding artifacts out of the rear and side surround speakers, yet the center and front LR speakers sounded just fine. I eventually found I could quickly determine which bitrates produced these artifacts by listening to the encoded song on headphones after opening it on CoolEdit and previewing the song using the channel mixer's "Vocal Cut" preset.
The "Vocal Cut" preset does a left minus right subtraction to the recording, which is similar to what Dolby ProLogicIIx does to stereo sources to produce surround information, and instantly highlights the perceptual coding shortcomings of using too low a bitrate.
Now, I used WinLame RC3 during my experiments, and it still has the option of choosing "Stereo" as well as "Joint Stereo" and "Force Mid/Side". Make sure you only use "Stereo" because the other two require higher bitrates to eliminate any encoding artifacts. (I originally thought the "Force Mid/Side" would work better since it already does a L-R & L+R function before encoding, but apparently it was designed to give more weight to the L+R content than the L-R information)
The settings I've empirically determined to sound indistinguishable to the raw wave file using the "Vocal Cut" method is:
- VBR with 160kHz minimum & 320kHz maximum.
- Stereo
- "Enforce minimum bitrate" checked off. (Mandatory!!!)
All the rest of the settings were at their default settings.
The end result is my songs now end up with an average bitrate between 180kbps and 230kbps, depending on how complex the song was. This provides a noticeable reduction of the disc space required compared to 320kbps, yet I find they sound identical to each other.
I haven't listened to the latest WinLame version so maybe these bitrates/settings could be reduced here or there to save even more disc space. Surprisingly, my son ripped a song from a CD using Itunes AAC encoding at 128kbps and it sounded indistinguishable from the CD's wave file using the "Vocal Cut" method. I'm convinced AAC encoding is definitely more efficient than mp3, but I'd gladly choose a format that's more common and not subject to any DRM to archive my collection.
Happy experimenting.
