Audio normalization

Combining audio files from multiple narrators requires normalization for consistency from slide to slide. Audacity and similar audio editors can normalize peaks but I've read that best results are obtained when the audio's RMS energy level is used for normalization, not the peaks.

What product(s) are eLearning developers using to normalize audio?

Sam

7 Replies
Sam Carter

I give Levelator high marks for simplicity. It does a good job but more than one 3rd party reviewer has indicated volume levels are sometimes inconsistent. Interesting the complaints come from MAC/Safari or MAC/Chrome users.

Could be a MAC issue, but I suspect the RMS calculations of the Levelator could be better.

Sam

Richard Lorrain

The problem with RMS is it may not identify the same average level on all files. Especially if you are mixing differents recorded sources. It may work well on longer files but in the end you might end up with 2 files RMS normalized with 2 differents global levels. I would go for peaking but look closely at peak levels close to zero; normalizing audio with peaks when you already have peaks to close to 0db might do nothing.  Also, before doing a peak normalization, I would strongly suggest an expander (see it as a reverse compressor) which expand the levels of your wave to express it more throughout the entire db structure. When you normalize audio, it tends to bring the floor level higher thus making it more noizy if the background noise is audible. 

Cheers!

Russ Sawchuk

Sam,

We often use two narrators speaking to each other in our courses. I tried "normalization" and other standards methods but none of them really did a good job in leveling the volume.

I finally found Auphonic Leveler and always use it on all of my audio!  Not only do I get consistent volume across slides, but also across all of my courses. It does a pretty good job in removing noise and hiss. You can try it free on their website.

Russ