Mp3: choosing the amount of compression

Question

I am on a custom-built pc, which includes 700 gigs of hard drives and 300 gigs of ram. I’m using Soundbooth 2.0.1 as part of the CS4 Master Collection, and I have high quality live audio recordings of readings of famous poems that I want to give away for free as downloads on one of my websites and provide a snippet to listen to on it. I want to save the edited Soundbooth  files as mp3s, but I’m lost when it comes to the amount of compression that I should choose. I’ve searched the literature and can’t find anything.
Does someone have a clue?
Karla

Answer 1

Here's a brief description of each of those functions, with some information on when you might use them.
Normalize:  Normalizing a file finds the loudest piece of the audio and adjusts the overall volume level so that this loudest piece is about as loud as it can get without clipping.  If this loudest bit was already clipping, then the volume of everything will be reduced.  This works well to boost levels as long as there aren't any parts that are much louder than the rest of the file, but if there is a big difference between the quieter and loudest parts of a file, it may not make much of a difference.
Hard Limit:  This boosts the volume (amplitude, actually) of an audio file, but prevent audio from clipping by setting a limit.  Any data that would exceed this limit is sort of squashed, so that if you overdo this tool, you can end up with distorted, muddy audio with little dynamic range.  It will be most useful when normalizing a file doesn't change the quieter passages much because of a spike or loud passage.
Soundbooth's "Louder" button:  The concept behind this tool was to address YOU!  Its purpose is to help make audio louder without clipping and with as little damage and degradation to the file as possible.  The first click performs a Normalize.  Not loud enough?  Click it again and it performs a Hard Limit amplification, to around -.3dB.  If there are still passages you feel are too quiet, you can continue to click it, though it might be damaging to the loudest parts of the file.
Honestly, as your dealing with recordings of spoken word poetry, there might not be a perfect, single solution.  Is the passage meant to have quiet, solemn passages followed by shouts of gusto?  In this case, performing too much volume correction would flatten these levels and reduce the impact.  Like a piece of classical music, it might have soothing passages followed by loud, climactic, powerful choruses before returning to sublime mellowness.  However, your intended playback device will be a factor as well.  Subtle dynamics sounds great on a large home stereo, but on a set of cheap laptop speakers, the listener would probably miss most of it.  If you're intended audience will likely be listening at their computers, on computer speakers or a laptop, then I would recommend foregoing the subtlety and making these recordings loud.
If the recording is entirely too quiet:  Normalize should help boost the level so that the height of the waveform, and therefore the amplitude level of the audio, is near the top and bottom of the display without going outside.
If the levels are wildly different, and it's apparent they were not supposed to be:  Try using Equalize Volume Levels to obtain a consistent level.  Then, if needed, use normalize or amplify to bring the level up.
You might also want to take a look at the "Match Volume" tool in the TASKS panel.  This will let you drag in several files at once and normalize them to either some average level - for instance, -3dB - or choose one of the files you're satisfied with as a Reference File, boosting all of the files to resemble that.