Posted April 4th, 2008, in: New Media| Technology| Web 2.0

After some experimentation, I found this is a good, quick, cheap way to get your audio levels for spoken word material up to snuff in a jiffy, without needing to understand much of anything about the technical aspect. Only requirement: a Mac.

Of course, all sources should be processed this way individually if possible.

If you’re recording Skype calls with CallRecorder, you should use Ecamm’s free tool (zip file download) for extracting the individual tracks from the saved Quicktime movie and import each side of the conversation into GarageBand on its own track.

Volume should be optimized for -0.5 db. Personally, I wouldn’t rely on an automatic process like Levelator for this. Maybe I’m old-school. I think it’s best to process each instrument/track with the following, in this order:
A. EQ (mainly to reduce rumble, plosives (‘P’ sound blasts), if there are any)
B. Compressor (Lessens the dynamic range of the material, allowing you to increase the volume with less clipping (overloading))
C. Limiter (takes care of the occasional clipping that occurs once you compress and boost the volume)
Then, put an additional Limiter on a Master Track, just in case.

Here’s how you could approach this using GarageBand:

Since GarageBand only has two open-ended plugin slots per instrument/track, you can take care of the EQ and Compression in one step (more or less), by using Apple’s built-in “AUMultibandCompressor” plugin (comes standard with every Mac). Assign the first plugin slot to “AUMultibandCompressor,” and the second slot to “AUPeakLimiter.” (pic)

A good place to start for setting up the Compression is the “Gentle” preset. Then, to get to the settings so you can fine-tune, click the little pencil button. First turn up the pre-gain volume until you have plenty of compression happening. You’ll know because all four of the meters on the bottom will be active almost all the time there is sound coming from that track (pic below).

Then, turn up the post-gain volume until the track’s meter (pic below) in the main GarageBand window is hitting the top on all the louder syllables, like the ones that start with P’s or K’s.

Do all this with the Tracks’ volume settings at their default positions. Only use the compression’s post-gain setting to increase the volume of the tracks, that way the Peak Limiter’s default settings will be in the right place. If the Limiter was set to stop the volume from going over -0.5db and you increased the tracks volume to +3db, the result would be that the audio could reach +2.5db, which is too loud. So leave the track volumes alone and only work with the gain controls in the settings for the Compressor plugin so that the Limiter’s -0.5 is the same as the Track Volume’s -0.5… (sorry if that’s confusing)

The goal is for the track volumes to get as high as possible without ever triggering the little virtual clip indicator lights (pic). If they do get set off, they reset by clicking on them.

If a track’s audio is hitting the red in the meter, but never tripping the clip indicators (pic), you’ve achieved the sweet spot of plenty loud, but not too loud).

Your podcast will seem as loud as everything else out there. Hurray!

NOTE:
If there is a lot of rumble from breath or microphone handling, or ‘Pops’ from P sounds (plosives), you can cut the Low-EQ of the track by reducing the “EQ 1″ setting of the compression plugin.
ANOTHER NOTE: If the sound starts to seem too artificial-sounding (squashed), back off on the pre-gain a little and compensate with the post-gain to get it back up to an adequate level (hitting the orange and red fairly frequently while never setting off the clip indicators)