Why be Normal? Understanding “Normalize” in your VO studio – Tuesday Tech Tip

Untangling the actual functionality of the most poorly named audio Effect…

“Normalize” seems to confuse more people in the voiceover studio than any other audio Effect available in recording software. Misunderstanding of the Normalize process regularly crops up in VO groups and discussions. I’m guessing there must be some resource which describes the Normalize effect in a confusing way. Every couple of months I encounter a new group of voice actors who are trying to make Normalize do something it cannot. 

We go through this tool in each of my recording classes – Recording at Home, Audition Ready Audio, and Post Production Polishing. But, it’s always worth hearing one more time to keep it clear, especially if you encounter someone who has it confused and you need to politely correct them. 

What Normalize does not do

The name does not help. “Normalize” does not make audio “Normal.”

That’s where the main misconception typically occurs. The name “Normalize” would incorrectly lead you to believe that it brings everything into the “middle” – that it makes loud things quiet and quiet things loud in the happy valley of VO sonic goodness.

Normalize does not do that.

Normalize will not change the tonal actual tonal balance of your audio. Normalize will not make things sound “different.” (More on that in a second…)

What Normalize does

Normalize is purely a change of amplitude. That means it is simply a volume (or “gain”) adjustment. Normalize has precisely the same effect as “Amplify” in your recording software. 

Normalize adjusts everything by the same amount. It just gets there a bit differently than Amplify. I’ll describe this a few different ways below.

Normalize will only make things sound “different” in that it can make it louder, and humans perceive changes in loudness as a positive. (it’s actually a bit more complex than that, as described by the Fletcher-Munson curve… but let’s not get distracted here).

The difference between Normalize and Amplify

It helps to think about them as the same process which gets there two different ways:

Amplify is a relative change of amplitude (and we perceive a change in amplitude as a change in volume). When we use Amplify, you’ll typically see a plus or minus adjustment in either dB (decibels) or percentage. Amplify doesn’t care where we start or where we end up. It only adds or subtracts a specific amount of Gain (making the waveform larger or smaller, which we “hear” as louder or quieter).

Normalize is a targeted change of amplitude. When we use Normalize, you will see a target value for either Peak (in dB) or overall Loudness (usually measured in RMS, LUFS, or LKFS). When you select a value for Peak, Normalize finds the highest Peak in your audio and does the math for you – taking the value of that Peak to precisely the targeted value. That means if you want to Peak Normalize to -3 dB, and the highest Peak in your raw audio is -13.753 dB, it does the math for you.

You could get to the exact same result if you simply used Amplify and added 10.753 dB to your audio.

The difference between Gain controls and Dynamics tools

If you want to adjust only certain parts of the audio, we need a dynamics tool, such as Compression or a Limiter.

Definition: Dynamic controls constrain the louder parts while letting us increase the overall volume. In other words, we are reducing the dynamic range (the difference between the quietest and loudest parts of your recording). In most applications, after we reduce the dynamic range, there is a “gain” step, which is why most people incorrectly think that compression makes things louder.

Generally, a dynamics tool is used when we master an audiobook recording and raise the RMS (average overall loudness) but keep the Peaks (loudest parts) under a certain level. 

With Compression or Limiting, we are reducing the dynamic range and amplifying the result.

With Amplify or Normalize, we are making everything louder or softer by the same amount. There is no “dynamics” control aspect with Normalize. If we use Normalize, everything (RMS and Peak) is changed by the same amount.

Here’s an example for dynamics: A classical symphony (Beethoven’s Ninth for example) typically has a wide dynamic range – there is significant change in the overall volume throughout the score. Punk rock does not – volume tends to be on or off, a low dynamic range though often high overall volume. 

A Couple More Notes on Normalizing vs. Amplifying
The difference between Normalizing and Amplifying has to do with how we apply the gain. When we use Amplify, we are adding or subtracting a specific amount of volume. Amplify does not care where we end up, only that we’ve changed things by a specific amount.

Normalize is “targeted”
We tell Normalize to change the audio to a certain specific value. If you “Peak” Normalize to -3 dB, the loudest single cycle in your recording is changed to that -3 dB target and the rest of the audio is changed by the same amount.  If the peak was at -12 dB, then everything would be increased by +9 dB. Normalize figures that math out for you. 

That’s why some of the online sites will give a simple recommendation to “Normalize to -3 dB” when they are just trying to encourage you to be competitively loud. 

However, that can cause another problem. Normalize finds the highest single peak in your file. If you had a single wave cycle at 0 dB (say something like a click you did not notice), then Normalize would actually reduce the overall audio volume by 3 dB (taking the 0 dB peak to -3 dB and everything else along with it. 

That’s why I encourage you to “cut down the tall trees” in your raw audio before using Normalize. One stray syllable can prevent you from being loud enough to be heard on your audition.

Another Way to Normalize
In some software (such as Twisted Wave), Normalize actually allows you to use other targets – such as RMS value (which is an average of the overall volume of audio). The danger here is that with RMS Normalizing, nothing prevents peaks from going above 0 dB. If that occurs, you may end up with distorted audio from that step, even if later on you reduce the volume. 

I’ll use RMS Normalizing when I want to compare audio – as I mentioned in this article – but I’m usually reducing the louder audio to match quieter levels, rather than boosting and potentially distorting. 

When I’m trying to boost RMS levels, I’m usually reaching for a transparent-sounding Limiter.  That is the tool which gives me gain adjustment while preventing the peaks from distorting. I suspect we’ll dig deeper into that topic before too long. 

Resource UPDATE: Tuesday Tech Tip: Thinking About Limiters

But for today, just remember than Normalize and Amplify are doing the same thing to your audio, just using different ways to get there.

Each week, I send a “Tuesday Tech Tip” to my email community. It includes technical tricks I’ve come across, refinements for voiceover workflow, and insights gained as a working voice actor and VO technical audio consultant. If you would like to receive these as soon as they come out, please take a moment to sign up here.