Professional Analog Mastering.
Professional Analog Mastering.
220K+ subscribers
Although each mix is a little different, the typical goal with backing vocals is to have them blend seamlessly with the lead vocal - in turn strengthening the lead, but in a way that isn’t obvious to the listener.
This means we want to control and attenuate anything about the backing vocals that makes them stick out, make them alter the perceived timing of the lead, or cause some form of dissonance with the lead.
The process for this can be a little time-consuming, but it isn’t too difficult if you follow the steps here and, of course, add your creative touches and use your discretion to help fit these ideas into your unique project.
For this session, I’m going to use a partially mixed instrumental stereo stem, then layer the lead and 2 instances of backing vocals so we get a good but straightforward representation of how to apply these ideas.
But if you have any additional questions that weren’t covered in enough depth, let me know in the comments.
Watch the video to learn more >
Although this chapter has to do with tracking and recording, it’s still good to keep in mind.
Sibilance or ess sounds, as well as hard consonants like T’s, K’s, and others, will make the backing vocal stick out.
When tracking, see if either you or the singer can attempt to lower the volume of these parts. One way is to sing these aspects softer; another is to move dynamically around the microphone.
To get a good example of how to do this well, I’d recommend checking out this video of Frank Sinatra during a studio session - I can’t play the audio for copyright reasons, but notice how he positions his mouth to create a balanced performance.
I’ll link the full video in the description if you want to check it out.But let’s go back to our session and talk about
Watch the video to learn more >
Editing is a powerful tool for getting backing vocals to sit well with the lead vocal - similar to recording the vocal properly; this will set us up for success when mixing and processing the vocals.
If we zoom in to the file, we’ll notice how the waveform corresponds to particular frequencies - for example, lower notes will be shown as a longer waveform, but sibilance and hard consonants are represented as a very clustered or condensed waveform. Since these parts of the performance are higher in frequency and usually have a high amplitude, they’re not too difficult to spot.
So once we find them, let’s isolate or split them into a separate clip, and then, we can use clip gain to attenuate them.
Routing-wise, these edits occur before any of our processing - so this will ensure that the track is balanced before we compress, de-ess, or anything like that.
Another more aggressive option is to move all of our isolated sibilance and hard consonants to a separate track. Then highlight all of them and turn them down with clip gain at the same time.
Then, highlight all of the remaining parts of the performance and normalize them. Each DAW is different, but in Logic Pro X, I’ll double-click one of the clips, open functions, select normalization, change the effect to ‘Individual Regions,’ change the algorithm to Loudness, and then normalize all of the regions to roughly -23 LUFS, which should be the default for this setting, at least in Logic.
Then, I’ll move all of the previously separated clips and put them back where they were. What we’ve done here is first attenuate the sibilance and consonants just like before but also create a very dynamically balanced performance for the rest of the backing vocal.
Just to clarify, though I’d never do something like this for the lead vocal, but on a backing vocal where we want it to be as uniform and controlled as possible, it’s a good option.
Let’s listen to the unedited backing vocals in the context of the full mix and then the edited backing vocals with both attenuated sibilance and normalized regions. Notice how the edited vocals sound much more uniform and are already blending in better with the lead.
Watch the video to learn more >
This next step may not be necessary if we edited well enough, but it’s still helpful to first insert a de-esser to at least check if the sibilance is still aggressive.
If you decide to edit, you can expect to achieve a few dB of compression with the de-esser. If not, more attenuation will be needed to remove the sibilance or at least make it less apparent.
I’d recommend using a split-band setting, and if it’s available, some look-ahead to ensure the de-esser captures the full transient of the sibilance or consonant.
On a lead vocal, aggressive settings like this would noticeably change the sound in an unwanted way, but since we’re blending these backing vocals in, we can be a lot more aggressive.
I’m using this FabFilter de-esser, but if you want a free option, your DAW should provide a stock option, or you can try this free T-De-Esser by Techivation.
Let’s listen to some additional attenuation to the consonants and sibilance and notice how it helps the backing vocals blend even better with the lead.
Watch the video to learn more >
For the 2nd insert, I want to equalize the vocal before I send it to a compressor or other processors - I’m going to use this Pro-Q 3, but your DAW should have a good stock alternative, or you could download this free MEqualizer by Melda Audio.
First, I’m going to introduce a high-pass filter to attenuate the vocal to right below its fundamental frequency. The fundamental frequency should show as a high amplitude spike, and it’ll be the lowest frequency shown by the analyzer.
Note that if you see some low amplitude signal below this fundamental, it’s likely a rumble from the mic stand, the recording environment, or plosives.
By attenuating up to the fundamental, we cut out any unnecessary or non-musical aspects of the recording. Then, I’m going to center a bell filter on the fundamental and attenuate it by 3-6dB.
Although this type of filter would likely be too aggressive for the lead vocal, on a backing vocal, we don’t need as much of this range since it’s already covered by the lead, and it’ll be a prominent range in the instrumental.
Next, I’m going to center a band around 500Hz and attenuate by a couple of dB. The reason is that this is where the majority of vowels are formed. Again, I don’t typically do this to a lead since I want to keep the vowel articulation intact, making it easier for listeners to pick up on the lyrics, but for backing vocals, it’s a less necessary range.
Up next, I’ll center a band around 2-5kHz - this is the area where we have the majority of a vocal’s clarity - if we were to amplify it, it would help the vocal stick out, but since that isn’t the goal, I’m going to attenuate the range by about 2-3dB, or as needed for the particular track.
If I’m still getting a good deal of sibilance, which shouldn’t be the case given our previous editing and processing, but if it is present, I could also use a narrow band to attenuate some of that between 6-10kHz. Use your ears to find the problem frequencies, but you can also observe where the sibilance is with the frequency analyzer to help you make a more informed decision.
Lastly, I’m going to set this EQ to a linear phase setting - this is kind of a strange choice, but let me explain real quick.
Linear phase filters introduce a mild form of distortion that primarily affects a signal’s transients. The greater the latency of the filter, the greater the effect on the transients. By enabling linear phase filtering on a backing vocal, I’m smearing the transients, so to speak, making the vocal sound less detailed, in turn, helping it blend in with the lead. Keep in mind that if you don’t have a linear phase EQ, there’s no need to worry about this part; just use a regular EQ.
Let’s take a listen to our 2 backing tracks in the context of the lead and mix and notice how we’re getting to a point where the backing vocals are only support for the lead.
Watch the video to learn more >
If we used loudness normalization during our editing, our backing vocals should already have their dynamics somewhat controlled. That said, more compression is usually a good idea.
I’m going to use this Pro-C2 compressor, but you could use a stock compressor or download the free MCompressor by Melda Audio if you need an alternative.
The first things I’m going to do are turn off auto-make-up gain, and enable lookahead. I don’t want the compressor to automatically amplify the signal in relation to how much it compressed since this wouldn’t reduce the dynamic range as much as I want - so that’s why this function should be turned off. Instead, I’ll manually compensate for the gain change once I figure out how much I’m going to attenuate.
As for lookahead, it’ll help the compressor attenuate peaks more accurately and without the need for a super fast attack time, which could distort the transient, making it slightly more noticeable.
When it comes to the compressor’s settings, a 4:1 ratio with a softer knee will work well. An attack time of about 10ms and a longer release will help attenuate the signal quickly and then smooth it out by holding onto the signal for a longer amount of time.
How much you compress will depend on the vocal performance, if you normalized the loudness of the track earlier, and how condensed you want the sound, but achieving a few dB of attenuation is a good idea. As I said a moment ago, I’ll manually adjust the makeup gain now that I know how much attenuation is being achieved.
So let’s introduce these settings to both of our backing tracks and take a listen to how they sound now.
Watch the video to learn more >
Up next, I’m going to tune the vocal. This is optional, of course - you might want to keep the vocal’s original imperfections, but I find it helps lessen how much I’ll have to tune the lead vocal.
Also, some engineers may want to insert the tuner earlier in the chain, but I like putting it after I’ve controlled the sibilance, frequency response, and control dynamic range. This way, the tuner is measuring a vocal with less unmusical aspects and with the performance dynamically consistent, resulting in better measurement and more accurate tuning.
Like some of the other forms of processing, I can be much more aggressive with my tuning than I would be with the lead. I’ll first need to find the vocal’s key - a service like TuneBat will work well for this. Just drag and drop either the lead or backing vocal into it to find the correct key.
Then I’ll set the tuner to the correct key and set the tuning to 100%. Faster settings will tuner more, but it may become too noticeable, so that a moderate setting will work well. Additionally, you’ll notice this particular plugin offers note stabilization, which causes the tuner to avoid quick note changes.
You’ll have to use your ears to find what’s best for your particular track, but I usually find this function to be helpful more often than not. Lastly, I like to avoid introducing doubling and stereo imaging with a tuner - some tuners have this option, some don’t, but I think it’s easier to do this type of processing with subsequent inserts.
Also, I’ve been using this Metatune plugin, but Melda Audio offers a good free alternative called MAutoPitch.
Let’s take a listen, and notice that since these backing tracks support the lead now, instead of standing out, the additional tuning makes it sound like the lead vocal is more in tune.
Watch the video to learn more >
Now that the vocals are controlled, balanced, and tuned, I think it’s a good idea to consider where they sit in the stereo image. You might want to keep them in the center, which is fine, but I usually find myself either panning, using stereo imaging, or a combination of the 2.
I do have a method that takes advantage of some psycho-acoustic ideas, and I think it results in really realistic stereo placement, so let me show you that in case it’s helpful.
Also you can use this method for other instruments, but let’s talk about it in the context of backing vocals for now.
The technique consists of 3 parts:
A Sample Delay plugin - usually, your DAW will have a stock option.
An EQ capable of affecting either the left or right channels - again, your DAW should have a stock EQ that lets you do this.
And then just the traditional pan-pot.
Starting with the sample delay plugin inserted after the previously discussed tuner, I’ll first decide if I want the signal to be on the left or right side. So let’s say I want this backing vocal to be on the left side.
What I’ll do is select the Millisecond option and then delay the Right channel by about 0.5 to 0.7ms.
Based on the speed of sound, this is how long it would take a sound wave to hit our right ear if the sound source was on our left.
This delay alone will place the vocal to the left side, but we can make it more realistic with EQ.
Next, we’ll insert the EQ and change the processing to the Right channel only. We’ll introduce a high-shelf filter and attenuate the high frequencies.
Since high-frequency waves are less powerful, our head will absorb them, causing the signal to have less high frequencies when it hits the ear opposite of the sound source.
Last up; we have traditional panning - this is really just altering the amplitude by attenuating the right channel if it’s panned to the left and vice versa.
If we subtly pan the vocal to the left, in turn reducing the amplitude on the right side, then the delay, the frequency response, and the amplitude collectively signal to our brains that the sound is on our left - resulting in super-realistic stereo placement.
Keep in mind that if you pan the signal all of the way, it’ll mess up this effect - we need some of the signal to still be on the right side so that we hear the delayed response and the attenuated high frequencies.
So let’s take a listen to this effect, with 1 backing vocal being sent to the left and one to the right, and notice how the placement sounds natural and realistic.
Watch the video to learn more >
Up next, I’m going to send any and all backing vocals, in this case just the 2 vocal tracks, to a collective bus by changing their outputs from the stereo output to an auxiliary track.
This will allow me to process them collectively.
On this auxiliary track, I can add any additional temporal processing that I feel is needed - be it delay, reverb, doubling, or anything that affects the time domain of the vocal.
Some quick reflections are usually great for doubling and creating a dense sound - or you could use some mild de-tuning and modulation for this, or you could introduce some tape slapback with a delay time of 160ms. Really up to you and the sound you’re trying to achieve.
Just know that if you add this type of doubling, there’s nothing wrong with following it with another, longer reverb if it helps blend the vocals better. I find that if the lead vocal has a long reverb, it helps to match that reverb on the backing vocals to make them sound more cohesive with the lead.
The same could be said about shorter settings, but you’ll need to use your ears and adjust for the situation.
One change I would make for any reverb you use is to lower the pre-delay - on your lead vocal, you‘ll likely want some pre-delay to include a short dry portion of the lead, keeping it from sounding washed out.
But on the backing vocals, having the full signal reverberated helps them blend in.
Like the reverb time or the type of temporal effect you pick, you’ll have to use your ears to determine how much of the effect to blend in.
For the demo, I’ll use this FabFilter reverb, but your DAW’s stock plugin is a great alternative.
Watch the video to learn more >
Last up for our backing vocal chain and mixing process, I’ll add a final EQ to the aux send - this way, I can equalize the vocals collectively and help shape them to fit better with the lead vocal and the instrumental mix.
I’m going to use this Sweet-spot EQ, which has very broad filters - making it great for general shaping. At this point, what I attenuate will depend heavily on the mix - maybe the highs are still sticking out too much, or I want the backing vocals to fill the mid-frequencies a bit more.
There are a lot of variables at this point that should influence your decision, but ultimately it’ll come down to what makes these backing vocals sit best in the mix, what helps them best support the lead, and what can I amplify or attenuate to balance the overall mix’s frequency response.
With that in mind, I don’t want to tell you what to do at this point, just that it’s helpful to have a final EQ to shape the backing vocals.
If you want a similar broad filter EQ but don’t have the one I’m using, try RareSE by Analog Obsession - it’s a really impressive free Pultec Emulation.
Let’s take a listen to the final sound and notice how the vocals sit comfortably in the mix, support and strengthen the lead vocal, and play a role in balancing the overall frequency response of the mix.