- by Kevin Deal
Audio normalization is when you put constant gain to a recording so that you can bring the average amplitude to the desired level. In other words, it means changing the audio’s overall volume to fixed amount. However, when doing so, you don’t change the signal-to-noise ratio or the relative dynamics in the audio. It just purely changes the volume. You can usually find this application in a digital audio workstation (DAW).
Importance of Audio Normalization
Audio normalization can be used when you want to get the maximum volume for your audio without having to change the dynamic range. This is specifically necessary when you come up with a somewhat quiet sound. If you just want to match the volumes of all your audio files or make them as close as possible, then you can normalize them too. This is usually useful with full mixes and snare hits. You can normalize an audio without changing the sound all together, unlike when you use compression. It gives you limited, but easy control over the audio.
Measuring the Audio Volume
There are several ways for you to measure audio volume. First, you can use the PEAK volume detection. It detects volume by looking at the loudness of the peaks of the sound’s waveforms. Some say that this is the best way to measure volume when you are thinking about increasing audio volume to a maximum since you can analyze how far it can go. If you study the waveforms, you will learn that you can’t set the volume higher than that of the highest peak.
Second is using RMS Volume Detection. Unlike the first one, this analyzes the overall loudness of the audio. It takes both the large and soft peaks into consideration and calculate the average. This is normally how human ear process sound so if you use this in your audio, you’ll get a more natural sound. However, RMS volume detection is not ideal when you want to master the audio. Moreover, if one track is consist of multiple frequencies, it will sound louder than what you set it to be.
Then there’s the EBU R-128 Volume Detection. This is a little similar to the previous one. However, it’s enhanced to emulate just like human ear. It can listen to an audio volume and analyzes how the humans perceive it. Unlike RMS, it takes into account the fact that humans can only hear frequencies from 1000 to 6000 Hz as loud.
Ways to Normalize Audio
Normalizing an audio usually depends on the kind of audio you want to produce. One way is to re-compress an audio which is already compressed. Of course, this will lead to the decline of your audio so only do this if you are sure that the track you’re editing is of high quality.
To begin, you should have something you can use to process the audio and give its levels a boost. There are free software you can get in the internet which can easily normalize levels. However, if you are not satisfied with the result, you might want to use a compressor or limiter to smoothen your audio.
If you don’t want to mess around with your audio, then it’s better if you use the appropriate hardware to maximize volume. You can buy an audio compressor and run your audio signal using it. When doing so, be sure to check its latency so you won’t get an audio that lags behind later on.
If the hardware is not enough, then you can use a software to do the job. Adjust the volume using the Sound Check feature. You can also download effect system, like Hear and Perian, to get the volume normalized.
Dangers of Audio Normalization
Peak normalization is when you have changed the gain to get the highest pulse-code modulation (PCM) value, which is the loudest level you can have in a digital system. If you change audio volume from peak normalization to 0dBFS, the audio may not clip automatically. This is inconvenient when you are using a multi-track recording. You can usually aid this by adding an extra processing or playing the tracks at the same time.
Overall, most audio engineers find normalization destructive as you can’t totally perform a digital processing to an audio without changing it. Even turning the audio up using peak normalization causes heavy strain to the sound. Therefore, you must only use normalization when necessary.