How To Make a Voice Sit Over Background Noise/Music

This tutorial has many applications. You could be trying to mix a singer's voice over a song, you could be talking over music in a podcast or radio package, you could even be doing the voice-over for a film. Regardless of what it is, you will face the same issue. The voice needs to be distinct over the background sound. There are many techniques to doing this, the most basic is to just dip the sound's volume when talking, but I'm going to show you some more advanced tricks for a more natural sound.

As the backing sound could be anything from music to industrial noise, I shall just refer to it as the "sound" or "background", it is all the same in the case of what we are doing.

The first part is picking a good background sound to go behind the voice. If you are recording vocals over a song, you already have this sound so you can skip this part. If not, there are many things you should consider. Let's use a Radio Package as an example, the reporter's voice is being recorded in the studio but they want the sounds from location added.

We want sounds that fit nicely around the reporter's voice, if someone is clearly talking in the background, this is not good as it will take attention away so either avoid voices or have indistinct crowd chatter.

This example is going to use motorway noise in the background:

If you listen to a motorway, it has sound from pretty much all frequencies. Imagine white noise, it sounds fairly similar. It also has a lot of voice-range frequencies which could mask the voice. So, the first thing we are going to want to do is identify the range of the voice.

Here is a screenshot from the EQ, it shows what frequencies my voice was hitting. We want these frequencies to be heard in the mix, so dip those frequencies out in the background noise's EQ.

This leaves room for the voice while still keeping all the other frequencies that identify it as a motorway. If you didn't know already, always try and remove frequencies instead of adding them when you EQ. The ear is less sensitive to missing frequencies and so removal is hard to identify. 

If you want to be clever, you could have the background noise duplicated, one with EQ and one without it. When the voice is speaking you crossfade over to the EQ'ed sound and when the voice isn't, you go back to the unedited track.

Now we can get onto some more tricks. The next thing I'm going to do will save you a lot of time. Many people automate their background noise to dip in volume while the person is speaking. This works fine but sounds very obvious. 

We are going to do the level up from that, side-chain compression. Probably the most talked about technique on this website, it has so many applications. Because this tutorial is not just for sound engineers and musicians, I will keep the descriptions simple. 

Sidechain compression dips the volume of the backing track when the voice speaks, in a natural way that flows along the contours of the volume.

A visual aid, this shows where the sidechain would dip 
the background noise in relation to the voice. This is,
however, very much oversimplified.

This means when the voice isn't talking, the background noise is clear and untouched, great for setting the scene or hearing the background music/sounds. But when the voice is talking, the sound dips accordingly. For full details on what it is and how to do it, click here.

A benefit of using sidechain compression is that when the voice is louder, the background noise will dip further, allowing for more room. This means it is less likely to clip as space is created relative to the volume.

If this concept scares you, you can always manually dip the volume relative to the peaks, however, you won't get as smooth an effect, and it will take ages.

If you are noticing that the voice is audible except for at the quiet parts, you should put a regular compressor on it. This will bring the volume of the quiet parts up. For simplicity, just flick through presets until you find one that sounds good or is appropriately named (e.g. male voice).