Improve Your eLearning Narration Audio With The Levelator

February 2, 2008

For my narrated eLearning courses, I do all my narration recording and editing in Apple’s fantastic iLife app, GarageBand. I plan to post a detailed tutorial of my GarageBand workflow in the near future for the benefit of Mac-based eLearning developers, as well as eLearning developers who are considering a switch to Mac. In the meantime, here’s a trick that may help you improve your own narration audio regardless of which audio application (or computing platform) you use.

To keep my recording process fast and efficient, I prefer to record my entire narration in a single continuous track. If I make a mistake, I simply repeat the sentence correctly and move on, without stopping the flow. Then I can make all my edits to the completed recording, split the track up by slide/screen, and export each as a separate audio file for importing into my eLearning development app. This may be the same approach you take when you do your own narration recordings.

The system is not foolproof, of course. There are elements beyond my control that force me to stop recording. Like airplane flyovers. The neighbor’s lawnmower. Kids knocking at the home office door (aargh). Lunchtime.

(Note: despite all these intrusions, in my case it’s still quieter and far more efficient than trying to do narration work at the office.)

This stopping and resuming of the narration recording can often result in uneven volume levels, as my mouth-to-mic distance may change, my mouth-to-mic angle may change, or I may resume my recording a little more enthusiastically (or, more likely, dispiritedly) than when I stopped. This vocal inconsistency is an issue I’ve found that is not usually resolved by standard normalization techniques. At least not with my lack of advanced audio engineering expertise.

Enter The Levelator.

The Levelator was developed specifically to resolve the problem of inconsistent volume levels in podcasts when you have multiple speakers 1) using separate microphones or 2) sitting at varying distances from a shared microphone.

So how does The Levelator differ from normalization?

Jon Udell explains:

In a talk given at the Podcast Academy, Daniel Steinberg noted that the normalization function in audio editors like Audacity and Audition “doesn’t do what you think it does.” I guess I’m not the only one who assumed, incorrectly, that normalization’s job was even out loudness variation. But apparently it doesn’t. It levels off peaks, but doesn’t raise up valleys.

Levelator does do what many people expect normalization to do. It brings the valleys up close to the peaks.

Like Jon, I always assumed that normalization helped even out audio across the spectrum - boosting the quiet spots and lowering the loud spots. Also like Jon, I was wrong in my assumption.

While The Levelator was designed for situations where you have multiple speakers or multiple mics, I also find it useful in correcting the inconsistent volume levels across my narration recording.

The caveat is that The Levelator doesn’t level out audio across multiple audio files. At least not as I understand it. So you need to use The Levelator on your narration recording before you start splitting it into separate slides/screens.

In GarageBand, I simply export my entire raw recording as a single AIFF file (the Mac equivalent of WAV). Then I run it through The Levelator and import it back into GarageBand, where I proceed to slice and dice it.

You can see the differences between the raw audio, the normalized audio, and the audio processed by The Levelator in GarageBand here:

On the first track you see the raw audio with its varying volume levels - there are loud spots and soft spots.

The second track has been normalized. You can see that the loud spots have been brought down, but the soft spots remain…soft.

The third track has been processed by The Levelator. The audio volumes have been leveled across the board. No more high spots, no more low spots.

Now, I’m no audio engineer, so I can’t tell you if there are any downsides to using The Levelator for this specific scenario, particularly when it’s doing its magic across sections of audio that don’t necessarily need any magic done.

But I do know that it’s been effective for me in eliminating the annoying problem of varying volume levels between sections of my narration recording.

Your mileage, as always, may vary. The good news is The Levelator is free, and available for Mac, Windows, and Linux. Give it a try and see if it helps your recording.

And feel free to share your results in the comments.

Side note: Articulate Presenter (my eLearning development tool of choice) offers an option to automatically adjust the volume levels of your slides to make them consistent (Optimize Audio Volume). I’ve used this option extensively in the past with varying degrees of success, probably because this option merely normalizes the audio.  And as explained above, normalizing may not be enough (hence my spotty results).

Posted in Audio