Audio Normalization: What Is It and Why Do We Do It?
First let’s define our terms so there’s no confusion. There are two kinds of audio normalization, and the first is Loudness Normalization. LUFS, which stands for Loudness Units relative to Full Scale, are a measurement of loudness over the entire length of an audio file — the podcast standard is -16 LUFS (recently challenged by Amazon and Spotify who suggest -14LUFS). There may be variation in loudness minute to minute but the average loudness is what LUFS is intended to reflect.
Loudness Normalization refers to the process of attaining this average. It is NOT what most podcasters mean when they talk about “normalizing the wave form”. That’s largely because Loudness Normalization is part of the Mastering process and most podcasters don’t have the tools or engineering know-how to truly Master their audio. Also, since most media hosts limit you to 128kbps mp3 files, many podcasters feel much of the Mastering process isn’t necessary. We disagree with that sentiment, for the record.
Peak Normalization is what most podcasters mean when they talk about normalizing. This is, in part, the process of ensuring that the loudest part of an audio file doesn’t exceed a certain db threshold — but that’s not all it does.
Let’s imagine, for the sake of an example, that we have an audio file whose loudest peak is -10db. Let’s further imagine you wanted the loudest peak to be closer to -5db. Setting a Normalizer to -5db and running it will increase the gain of the loudest peak from -10db to -5db for a gain of 5db. It will also increase every other peak by 5db. -15db is now -10db, -12db is now -7db, and so on. Peak Normalization is the automated and even-increase of db across an entire audio file by a fixed amount.
Why do we normalize audio in these ways?
Loudness Normalization is intended to ensure that, episode to episode, the average volume of your program is the same. If every episode has its loudness normalized to -16LUFS, no one will ever have the jarring experience of going from episode 13 to episode 14 and having their ear drums blasted out because this week’s episode is twice as loud as last week’s. We (the Portland Pod) also do this because it’s an audio standard and we like standards. ❤
Peak Normalization is intended to ensure that no single wave peak is loud enough to offend the ears of your listeners so badly that they rip their earbuds out and throw them ground while screaming,
“I will never forgive this podcast! I’m going to write them a terrible review!”
I know you’re wondering, because I was too when I was first learning about this stuff,
“But if we’ve taken care of the average loudness, haven’t we already prevented the problem of hurting our listeners ears?”
And the answer is no. You could easily have a half-a-second audio peak that kills your listeners ears without knocking your LUFS number down to less than the average you were going for — remember, LUFS is over the long-term, and averages are a poor measurement of quality when used on their own because averages don’t reflect exceptions.
If you have questions, feel free to ask! Hope this was helpful.
From the comments: reader James Cridland, Radio Futurologist and Editor of PodNews, suggested the following article of his for a more in depth look at what I’ve outlined above. Have a read and be sure to subscribe to PodNews — it’s a great daily read full of podcast-related must-knows and news. Read the article here: https://podnews.net/article/lufs-lkfs-for-podcasters