Professional Mastering - Analog & Digital
Professional Mastering
251K+ subscribers
So, you’ve recorded, edited, and mixed your vocal, as well as made it work within the context of the full production.
But, something is still slightly off.
Just about every mixing engineer and producer has experienced this, and it’s frustrating to say the least.
But there are some techniques I started using a few years back that solve this problem every time I encounter it. I’m going to share those with you today and demonstrate everything step by step so that you can stop having to revisit a mix over and over, as this problem often causes.
So, let’s jump into The Final Stages of Vocal Mixing - How to Compete with a Pro Vocal Mix
Before we get into the first technique, keep in mind that these techniques occur near the end of mixing a vocal. Your EQ, compression, de-essing, reverb, delay, and so on have all been introduced by the time you introduce these ideas, and you just need those last few small changes to complete the sound.
Watch the video to learn more >
My favorite part about making these videos is that I consistently learn new things and explore new audio topics - when making a video a few years ago, I came across this research paper titled ‘Intelligent Audio Production Strategies Informed by Best Practices.’
This thing is a treasure trove of fantastic info - but perhaps the most impactful sentence in this entire thing is:
“We have concluded that on average the vocals are equally loud to the sum of all other tracks…”
In other words, in the 928 professionally mixed songs they studied, the vast majority set the lead vocal and its doubles to the same loudness as all other instrumentation.
For example, if your mix is just a vocal and an acoustic guitar, they’d be equal in loudness.
Or, if your mix has a vocal, drums, bass, and guitar, then the vocal will be as loud as the drums, bass, and guitar combined.
To check this in your mix, send the vocal, its doubles if you recorded them, and all vocal sends, like vocal reverb or whatever you’ve set up, to one bus. Then send everything else to another bus.
This way, you have your full instrumental and your full lead vocal isolated to two channels. Then, add an LUFS meter to each, and compare their integrated loudnesses.
If they’re the same loudness, then you’ve established the most common and standard vocal loudness. If not, you can choose to adjust as needed. The genre may call for a slightly louder or quieter vocal - but, now you know the standard and can use it moving forward.
If you’ve ever had a vocal that sounds perfect during a sparse section of the song, but then gets buried during a chorus, even if you increase its volume, this works incredibly well.
There are multiple ways to do this, but I’ll show you what I’ve found to be the most effective.
In short, I’ll use a dynamic EQ to subtly expand the vocal’s third formant whenever that range is loud enough in the instrumental.
You’ll need a dynamic EQ that lets you use an external sidechain as the band’s trigger.
Alternatively, you could use a multiband dynamics processor - again, so long as you can trigger the band with the external side chain.
I’m using this Kirchhoff EQ, but use whatever works for you - I’ll insert the EQ on the vocal. Then, I’ll side-chain the instrumental bus. To simplify things visually, I’ve bounced down our two buses to an instrumental and a vocal track.
I’ll create a bell that’s centered on the vocal’s 3rd formant - this is a cluster of higher frequencies, typically between 2-5kHz, that’s responsible for what we percieve of as clarity.
I’ll make that band dynamic and create a positive range for whenever the signal crosses the threshold. For detection, I’ll set the trigger as the external side-chain’s mid image - again, this external side-chain is the instrumental track.
I’ll then determine the frequency bandwidth of the instrumental that’s measured. I’ll set this bandwidth as slightly smaller than the bell’s bandwidth, and with a slightly lower center frequency.
While playing both, I’ll adjust the threshold and ratio until I’m getting dynamic expansion to the vocal’s formant range.
So what are we doing here, and how does it help?
In short, whenever more of the instrumental enters, like during a chorus or busier part of a song, the overtones of these new instruments occupy the mid-frequency range.
This increases the amplitude of the mid-frequency range, which in turn masks the higher frequencies of the vocal, primarily the third vocal formant.
When this formant is masked, the vocal is harder to hear and sounds less clear, or slightly muddled.
This is why it can sound great during a less busy part of the song, but gets buried when new instrumentation enters.By amplifying the vocal’s 3rd formant when additional instrumentation enters, we reduce masking and retain the vocal’s clarity during busier sections of the song.
If you try this, I’d recommend experimenting with different center frequencies for both the dynamic bell and the detection range.
Additionally, if it’s an option, you may want to isolate the detection to the instrument’s mid image, since its side image will have much less of a masking effect on the centered vocal.
Let’s try this out, and let me know if the vocal retains its clarity when more instrumentation enters.
Watch the video to learn more >
I’ve discussed critical bands a few times before, so I’ll be quick - in short, a critical band is a bandwidth of frequencies that are easier for our ear to distinguish from other bands.
This is due to the shape of the cochlea, how the cilia or auditory hair cells interact with sound, and other factors, but the main takeaway is that research shows we have 24 critical bandwidths.
By amplifying the center frequency of these bands, usually with a 1/3rd octave filter to reduce band overlap, we can greatly reduce masking.
I have been using this Eqivocate plugin since it has a critical bands or bark bands preset - the filters are also 1/3rd octaves and use a unique pointed peak filter to really emphasize the center frequency, but use whichever EQ you’d like.
Maybe pause the video and screenshot the frequencies shown here, and recreate the filters and center frequencies in the plugin of your choice.
Anyway, I’ll listen intently and then begin to amplify and attenuate bands as needed.
Usually, 250Hz can come down a little, while the area around the 3rd vocal formant of 2kHz-5kHz can come up, but as I’m sure you know, this depends heavily on the vocal performance.
Let’s listen to a vocal set to the same LUFS as the instrumental, with the frequency specific side chain setting used, and lastly, with the critical bands adjusted.
Notice how these final steps really help the vocal sound polished and finalized.