Equalization for Voice Over and Audiobook Narration

Frequencies of speech

Human hearing is able to hear sounds in the range between 20 Hz and 20,000 Hz, dependent on age, health, and luck. As we age, the higher frequencies just disappear, and the lower frequencies sound muddled.

The range of frequencies required to reliably interpret spoken words is relatively narrow. In fact, older telephone technology was restricted to a narrow frequency range of 300–3,300 Hz, called the voiceband¹.

Of course, technology has advanced enough to allow more faithful reproductions of sound, and a wider frequency range is more pleasing to hear. There is, however, a point where too much range becomes detrimental to the clarity and enjoyment of the human voice, as heard on a recording.

Cleaning up the sound

Using Reaper, I suggest bussing all your vocal tracks to a single vocal track. That way you can apply effects once, and have them apply to all relevant tracks. Then select that master vocal track and add an instance of ReaEQ, the built-in equalization (EQ) FX. Like most effects built into Reaper, it is high-performance, and has low system requirements.

The image below shows what you should aim for as your initial EQ pass. It has steep dropoffs just below 80 Hz and just above 16 kHz. Why? Keep reading!

High-pass

Recorded spoken sound lower than around 80-100Hz does add some character to a voice, but is detrimental to clarity. You remove them in your EQ with a high-pass filter. Just think of it as allowing the high frequencies to pass.

With ReaEQ open, double-click along the central horizontal line, and drag the resulting marker to the left until it reads 80 Hz. For men, I do not recommending going much higher, but most women can go a little higher, perhaps up to 100 Hz. You can experiment with your own voice to get a feel.

Change the EQ type (far left, second down from the top) and change it to Low Cut (Butterworth). This makes the EQ have a steep drop-off lower than the placed marker. Selecting just Low Cut changes the shape of the transition as it drops off, making a smoother drop-off, but will may be audible which is not really desireable.

Change the dB of the cut to the maximum available, 120 dB. We have no need of any sound below this point.

Low-pass

Recorded spoken sound above 16kHz adds some sparkle to a singing voice, but sounds in that range are often tiring to the listener over long durations. You remove these in your EQ with a low-pass filter. Remember the name by thinking of it as allowing the low frequencies to pass.

Similar to what was done for the high pass, double-click along the central horizontal line, and drag the resulting marker to the right until it reads 16,000 Hz. For men, I do not recommend going much higher, but some women may benefit from going a bit higher. Listen to your voice as you move the marker higher or lower. You are listening for changes that affect comprehension and character. If it truly does not add to the value of your voice, trim it!

Quality vs. Latency

As with most things, there is a tradeoff between high quality and latency (speed). ReaEQ has a setting that allows you to select which of those two factors you want to prioritize. Along the bottom row, at the far left, click the bubble labeled HQ, and you will see two options: “Economy - Fastest, No latency” and “High Quality - Accurate, Latency”.

If you were to use this EQ in a live-monitored setting, such as if performing in front of a live audience or broadcasting, you would not want any latency introduced. Listening to the sound of your voice on your monitor headphones might be delayed slightly, and would be distracting.

Voice over and audiobook narration are not usually performed live (thankfully!), and therefore I recommend you set this configuration option to High Quality.

Final thoughts

Removing those frequencies first, before any other processing or adjustments, should be your first step. When you listen back to your recording you will hear only the frequencies you intend to keep, and any further effects you add to your mastering chain will not be impacted by these garbage frequencies.

You will also notice that some of your room noise will be attenuated or reduced. The sound of your furnace and sometimes your computer or laptop fan might be reduced. A forthcoming post on taming your noise floor will discuss this option and more.