This post is going to cover some basic points about how to mix vocals and the frequency range of the human voice. We'll look at equalising vocals, whether those are spoken or sung.
Let's assume that you at least have a reasonable vocal track. The key to good recording is to get the best original signal possible. We're not discussing recording here, just equalising the track after it is recorded. We're not accounting for other noises that could be embedded in your take such as wind, traffic rumble or room noise. Those can be minimised to some extent. As we just mentioned the key is getting the cleanest take possible.
We're using screenshots from some of the more common software that people to use to record and process audio. There's both graphic and parametric equalisers.
Let's start at the beginning. What are Hertz? What frequencies can the human ear perceive, and where does the human voice fall into that range?
The term Hertz came about through sound engineers wanting a more specific term than, “those sound wave thingys.” Hertz basically means how many cycles of a soundwave occur each second. Low notes have a lower number of cycles per second and high notes have a high number of cycles per second. The human ear can perceive from 20Hz (or cycles per second) to 20,000Hz. You can see it represented in the image below.
Ignore decibels for this entire article, we're only dealing with frequencies.
In the image you can see the typical range for average male voices and for female voices. Not everybody has the same range. Some people have lower voices, others higher, but they will usually fall somewhere in those given ranges. The range of human voice is about 80Hz to 4kHz (kiloHertz), with the upper range cited as anywhere between 3kHz and 5kHz. Most often, 3.5kHz is given as the upper vocal register. The difference in stating the approximate frequency can vary on how that upper register is defined; for example, functional range vs possible range.
So, does that mean when equalizing the human voice we strip off everything outside that range, like is shown in the diagram? Or if using a male voice, for example, rip out every frequency above 900Hz. Not exactly.
Let's take a quick look at soundwaves. Here we have a diagram of a 440Hz soundwave. That's the note of concert pitch! In standard tuning your tuner will be set to 440Hz. It's the A4 note on a piano keyboard. This is the pitch that was decided every instrument should tune to. Of course, there are higher and lower tunings, but this is the note for 'standard' or 'concert' tuning.
The next image shows 1kHz (1,000 Hertz). You can visually see that there are more cycles per second in the higher note.
The thing is that sounds almost always have 'harmonics' or undertones embedded in this. These harmonics are an essential part of defining the sound, tone, or timbre of a noise such as a human voice or musical instrument. Soundwaves travel in both directions upwards and downwards in cycles. As mentioned humans detect soundwaves between 20Hz and 20kHz. You probably know that animals such as dogs and bats can perceive much higher frequencies than we can hear. Other animals, such as elephants, perceive sounds lower than we can hear.
The harmonics of any given frequencies are cycles that match in halves or doubling the main signal. In the diagram below we've superimposed 500Hz over 1,000Hz. You can see a pattern of matching. Those patterns will also match again with 2kHz (2,000Hz) and onwards and upwards, and also downwards. Our 'concert' pitch of 440Hz will also have corresponding cycles travelling in both directions 220Hz, 880Hz, and so on.
The argument is that if we simply cut off the frequencies that are not needed then we're loosing some of the original signal plus it's harmonic overtones. The general solution is to 'roll off' what we don't need.
Equalizing becomes somewhat of an artform. You could imagine then cutting off all the frequencies in a band mix consisting of maybe 22 tracks, so that each one sits nicely could be a challenge. Which frequencies to 'roll off' and which ones to just drop. It's slightly off topic, but a good approach here is to single out which are the most important instruments in a track, 'vocals & keyboards?' Your kit and bass guitar will also usually be processed in relation to each other, holding down the low end.
Right... back to vocals. Here's a guide to equalising vocals.
<80Hz – Roll off these frequencies unless they are helping the mix.
100Hz – 350Hz – you get a lot of 'boom' in this area. Try not to boost it, especially on male voices.
350Hz – 500Hz – a lot of body is in these areas. Remove the exact frequency problem with a very tight drop. Don't use wide attenuation or you may make the vocals sound weak and hollow.
800Hz – 1.5kHz – attenuate problem areas here by cutting them out.
>4kHz – increase slightly if needed to provide clarity but avoid increasing unwanted sounds.
>10kHz – Exercise care here, see tips below.
A few more things to consider:
- The range of a male singing voice is considered at its lowest note, a 'Bass' voice, to be 82Hz, and the highest note of a 'Soprano' voice, 1050Hz.
- Sibilance is a hissing 'Ssss' sound. It occurs in the higher frequencies. If you're boosting vocals in this area to increase presence and clarity be careful you don't bring in sibilance.
- Above 10kHz humans have less aural sensitivity. That is, we don't hear as good in this higher area, the more so as we age.
- Roll off the low frequencies, but not so much as to loose presence in that region. (or you can increase it to create 'boom').
- 3kHz to 6kHz adds clarity and brightness.
- 4.5kHz to 6kHz adds presence.
There are no hard and fast rules. Each track is unique. The tips given here are a good guide, but you'll need to work with your track to bring out its best. Close your eyes and listen. Does it sound natural? Is it clear?
As for equalisers, basically a graphic equaliser works in specific and preset bands. A parametric equaliser has more options that allow the frequencies of those bands to be adjusted and how wide each band is.
Typically, studio engineers will create a very tight band on a parametric equaliser, push it up to maximum and then sweep through the frequency range of a track to find audio problems. They then strip that exact frequency. This technique also works for live music. One solution to feedback problems can to be to find the exact frequency that is causing the feedback and strip it out entirely whether using graphic or parametric equipment. You can see this technique illustrated in the diagrams below.
Lastly, have fun experimenting with this sine wave generator. We discovered it at Academo. See for yourself the difference in sine waves between high and low frequencies.