Perfect Vocals – how to get Rid of “Boxiness” Using Multi-band Compression
We’ve all been there. You’ve set up your own home studio and recorded what you thought was a great vocal take. But when you listen back later, you find the vocals sound “boxy.” Why might this happen – and how can you use multiband compression to fix it?
First, let me define what I mean by “boxy vocals.”
Terms like “boxy,” “muddy,” or “boomy” (or even “honky”) are subjective, and whilst they might mean slightly different things to different people, they broadly cover the same issue – having excessive energy in any one particular frequency range. To me, if a vocal sounds boxy, it means the vocals feel suffocating. Or, to put it another way, the vocals have a lot of energy in the low-mids, somewhere in the region of 150hz – 1Khz.
Causes of Boxy Vocals
Vocals can often sound boxy because, in effect, they have been recorded in a large box. Vocals are often recorded in vocal booths or small rooms that have poor acoustic treatment – or aren’t treated at all. Adding thin foam tiles to the walls only reduces the highest frequencies, leaving the lows and low-mids untouched (and often exaggerated due to the dimensions of most home studio vocal booths).
Even in larger, well-treated rooms, the proximity effect (being close to the mic) can mean that low frequencies are exaggerated. This is fine – or even desirable – for a radio DJ, but it’s far from ideal in most mixing scenarios.
Whatever the reason, there is thankfully a solution – which I’ll describe below.
It should be worth noting that the techniques don’t have to be reserved for particularly problematic vocals. In fact, I personally do this on almost every vocal I mix in varying amounts, depending on what’s suitable.
Solution to Boxy Vocals
The issue with achieving a balanced vocal sound is that a vocal performance is often quite dynamic, and the singer can move through different registers and use different singing techniques (think singing from the chest vs. notes that are more breathy or falsetto), all in a short space of time.
An EQ setting might be suited to one section of the vocal, but not the next. For example, cutting the low mids might feel suitable when the vocalist is singing in a lower register, but the same EQ setting might make the vocal sound thin when the vocalist is singing in a higher register.
The images below are taken from the same vocal line, moments apart. You can see the resonant frequencies that make the vocal sound boxy are much more pronounced in the first image when the vocalist is singing lower notes. I’ve included audio examples at the end of the article – listen for the difference in the words “I know” vs. “that you said” in the first line.
The solution, then, is to control any particular frequency area dynamically, rather than statically as you would be with a traditional EQ. In other words, you need to be able to react to the audio and only reduce a frequency range when actually needed.
You can do this with a multiband compressor or dynamic EQ. There are some differences between the two devices, but you can achieve similar results with either. There are also other tools that can be used, which I’ll mention later in the article.
Vocal Sound Settings
The idea here is to reduce a frequency range, but only when it becomes too loud. So, how do you do this? In the case of a multiband compressor, first, find the area you want to focus on. This will be somewhere in the region of 150hz – 1khz. If your chosen device has a spectrum analyzer, you might be able to see the problematic area. Then, adjust the threshold until you see some gain reduction. It might be worth looping a section of vocal where boxiness feels particularly pronounced, then making sure the settings are right here.
How much gain reduction you will need will depend on the vocal recording. The aim isn’t to attempt to remove this area entirely, as doing so can sound unnatural or leave the vocal sounding thin or weak. This area also can contribute to “warmth,” so use your best judgment and aim for a sound that feels more balanced so that the problematic notes are no longer “jumping out.”
If your device has a ratio (not all of them do), try starting at 4:1. Then, depending on what’s needed, you can be more or less heavy-handed.
The images below show the compressor is only reducing the low-mids when necessary.
Attack and Release
The magic of this technique lies in the attack and release settings of the multiband compressor. The attack can be set up to allow just a hint of the low-mids to sneak through before the compression grabs them and reduces the volume. Doing this means the vocal still sounds full, but because those frequencies are only audible for a short period of time before they are controlled, they don’t hang around and interfere with your mix. The exact attack settings will depend on the vocal, but you can try starting around 8-10ms and work from there.
Note – If the vocal is particularly problematic in this area, and you want as little as possible to sneak through, then using a shorter attack time may be more suitable.
For the release, go for a time that’s quick enough to allow the compression to reset during pauses in the vocal, but not so quick it sounds unnatural or introduces “pumping.” This really does depend on the rhythm and tempo of the vocal. The term “use your ears” comes to mind, but this isn’t always that helpful if you’re new to a particular concept. Most plug-ins have some sort of visual representation, so use that as a guide and aim to have the compression resetting and “dancing” sympathetically to the groove of your vocal.
Other Uses for Fixing Vocals
This technique isn’t reserved for just the low-mids. You might want to try a similar approach on the lowest frequencies. Often you will have already high-passed the vocal track – for example, to remove unwanted plosives and rumble – but I regularly hear mixes where people have gone way too high when cutting low frequencies (perhaps to combat this issue). Being less aggressive with high-pass filtering and using this technique instead should lead to better results, especially if it’s just to combat the odd problematic notes.
You can also use it anywhere else in the frequency spectrum, of course. If the vocal is a little harsh around, say 2khz, but only when the singer hits certain notes, then try this technique instead.
Multiband compression is useful on many other sources too, not just vocals – the lower notes on an acoustic guitar, a resonant synth patch, or even a static sampled snare. You might wonder why this would be useful on a static snare that doesn’t change over the course of a song. Still, you could, for example, retain the transient thud while reducing any sustained lower frequencies as the sample decays – being able to control frequency content over time is a seriously powerful tool.
Tools to Fix Boxy Vocals
There are plenty of plug-ins that should be suitable for the job, and I’ve included the list below. It’s not an exhaustive list, and I can’t say I have experience with them all, but most multiband compressor or dynamic EQ plug-ins should get the job done – as long as you can set the frequency bands yourself (Xfer Record’s free OTT multiband compressor, for example, has set frequency bands that can’t be changed).
- FabFilter – Pro-MB or Pro-Q 3*
- Waves – C4, or C6 or F6*
- Izotope – Ozone 9 Advanced (includes both multiband compression and dynamic EQ)
- Universal Audio – Precision Multiband (comes free with some UAD interfaces)
- IK Multimedia T-RackS – Quad Comp
- Blue Cat – MB-5 Dynamix
- Softube – Drawmer 1973
- Vengeance Sound – Multiband Compressor
- McDSP – ML4000
- Melda Production – MDynamicsMB
- Stock plug-ins – don’t forget to check the plug-ins that come bundled with your DAW. You might well already have what you need!
When using a dynamic EQ, such as the Pro-Q 3 from FabFilter, you may not have options for attack and release. The Pro-Q 3 in dynamic mode (right-click on the EQ node and select “make dynamic”) is program-dependent, meaning you don’t set these values yourself, but the software is carefully tuned for optimal results depending on frequency range and dynamic range. It usually does a great job too, so this may be a better option if you’re just starting out.
There are some newer plug-ins that I’ve listed below that also do a great job. My workflow includes using these more often than multiband compression now, but the concept remains broadly the same.
Oeksound – Soothe
While Soothe has been around for some years now, it’s still a comparatively new concept. Soothe is a “dynamic resonance suppressor.” You boost the detection circuit in the area you’d like to focus on, and Soothe reduces resonant peaks in that area. You still have control over attack and release, but it can be more useful than multiband compression, as it doesn’t affect nearby frequency areas. It can track problem notes with extreme accuracy.
Eventide – SplitEQ
Split EQ is a new concept for a single plug-in and differs from a traditional EQ, as you can treat transients and tonal information separately. This means you could retain transients in the boxy area but reduce just the tonal information.
I have included audio examples of the technique in action from a mix/production project I’m currently working on for a client. I was sent these vocals recently, and it happened to be a great example for this article. While the vocals are generally well recorded, there are obvious resonances in the low mids, causing the vocal to feel quite claustrophobic. Notice how the high frequencies become more apparent – it almost feels like these have been boosted, when actually I haven’t done anything to them. I haven’t even increased the volume of them to compensate. Generally, you should find, by controlling the low-mids appropriately, there is added clarity and presence without boosting the top end.
If you’re newer to mixing, it’s often tempting to boost the high end of your vocals if they feel boxy or muddy, but this can lead to a skewed tonal balance that’s difficult to sit in a mix. Hopefully, this demonstrates why paying attention to the low-mids first is a sensible idea – you can, of course, always still boast the high frequencies (or turn the whole thing up) if further presence is needed, but you’ll be in a much better position to do so by looking at it this way round.
In the Mix
High pass filter, light compression, light de-essing, send FX, no EQ.
With multiband compression.
Without multiband compression.
High pass filter, no other processing.
With multiband compression.
Without multiband compression.
About Paul Aspin
Paul is a mixer, producer, and business owner who spends his time working on projects for clients, producing his own music, and teaching others to do the same. If you would like to take 1-2-1 lessons with Paul or hire him for mixing, please get in touch via the website or via e-mail:
Website (lessons): www.fullfatmusic.com
E-mail (mixing and general inquiries): email@example.com
Audio Production, Keeping Track