Getting a song to sound professional comes down to two things: how loud each part is, and how the EQ fits together. Nail that balance across your vocals, music, drums and bass and the track works. Miss it and you can spend hours fighting the mix. This guide is how I get both right across a whole mix.
Set the volume balance first, because a lot of mixes that aren't working have a level problem rather than a tone problem. Then EQ every channel in context, against the rest of the mix and against a record you trust, with every move level matched so you hear tone instead of loudness. Work on your vocal, music, drum and bass buses together, and shape the center and the sides separately when one stereo move would fix one thing and break another.
The rest of this guide covers each step, with audio you can hear, and shows how CARBON Q does the volume balance and the EQ of your whole session from one window.
Most of the mixes I'm sent for mastering don't have a processing problem. They have a balance problem. The vocal is a couple of dB too quiet, the bass is fighting the kick, the guitars are sitting on top of the vocal's presence range. None of that is fixed by a better plugin. It's fixed by getting the levels and the tonal relationships between the parts right.
This page is everything I know about that job in one place. Each section answers one question and links to a deeper guide when you want more. Read it top to bottom, or jump to the part you're stuck on.
- Why balance is the mix
- Get the volume balance right first
- Every EQ move is relative
- What each frequency range does
- Level match every move
- EQ the whole mix at once
- Borrow the balance of a record you love
- Smart starting points
- Mid/side across the whole mix
- Holding the balance through loud sections
- Clean up before you balance
- The workflow, start to finish
Hear the difference
The same rock mix with CARBON Q off and on.
Rock Mix
1Why is volume balance the mix?
Play a record you love through a phone speaker. A lot of the detail is gone and the low end has disappeared, but it still sounds like a record. Play a rough mix through the same speaker and it falls apart.
That's almost never down to the processing. It's the balance. Your ear picks up the relationships between the parts before anything else. Is the vocal in front of the music, or behind it? Is the kick working with the bass, or fighting it? Get those relationships right and a fairly simple mix sounds professional. Get them wrong and you'll be fighting the mix for hours.
That figure is the whole idea in four faders. The vocal comes up 3 dB, the music comes down 2, the bass comes up 2, and the mix goes from off balance to in balance without a single EQ move. The full method for getting there is in how to balance all the elements in a mix, and if the vocal is the part that never sits, read the perfect vocal loudness (with data).
2Why should I set the volume balance before I EQ?
The pros give the volume balance the attention it deserves. Beginners skip it because it feels too simple to be the answer. Opening an EQ feels like real engineering. Moving a fader feels like you haven't started yet.
I've fallen for this loads of times. The vocal isn't sitting, so I try to fix it with EQ. Cut some low mids. Add a bit of presence. Maybe a touch of saturation to help it cut through. Twenty minutes later it's brighter, a bit thinner, and still not working. I bypass the EQ, turn the vocal up a bit, and it works straight away.
EQ and volume are tangled together, which is what makes this so confusing. Change a channel's level and its tone seems to move with it. Change its tone and it seems to change level. It's probably the biggest time sink in music production, and the reason a lot of people go round in circles for an evening.
So check the balance first. If the problem is level, EQ won't fix it. If the balance is right, the EQ moves you do make will be smaller and much easier to judge. There's more on ordering your moves in prioritizing your EQ moves for maximum efficiency, and on vocals that won't sit in balancing vocals in a dense mix and how to make a vocal cut through without pushing the fader.
3Why is every EQ move relative?
Solo a vocal, EQ it until it sounds amazing, drop it back in the mix and it doesn't fit. We've all done it. The vocal was never the problem. A channel isn't bright or heavy or present on its own. It's only those things compared to what's around it. Brighten the vocal and everything else just got darker.
So most tonal problems are really two channels wanting the same part of the spectrum. Your guitars and your vocal both sitting at 2 to 3 kHz. Your bass and your kick both filling 80 Hz. Solo either one and it sounds fine. Together they clash, and the usual reaction is to push the one you care about louder, which makes it worse.
It helps to think of the spectrum as a budget. There's only so much room at the top. Lift the highs on your vocal, your cymbals, your guitars and your synths and you don't get four bright elements. You get a harsh record. Same at the bottom: a big low end on the bass plus a big low end on the kick adds up to mud. Records that work have already made those trade-offs.
Decide who owns each zone. The lead vocal usually owns the presence range, so make a gentle dip in the guitars and keys around 2 to 4 kHz. Either the kick or the bass owns the sub, and the other one gets a pocket. I go deeper on this in 7 ways to fix (or even prevent) frequency collisions, creating space in your mix using EQ and how to get a great tonal balance in your mix.
4What does each frequency range do?
To share the spectrum out, you need a map of it. The audible range runs from about 20 Hz to 20 kHz, and it splits into zones that each do a different job. When someone says a mix sounds muddy, boxy, harsh or thin, they're describing too much or too little in one of those zones.
- Muddy: too much at 120–250 Hz, usually stacked up across lots of channels.
- Boxy: too much at 250–500 Hz.
- Harsh: too much at 2–5 kHz, where our ears are most sensitive.
- Sibilant: spikes at 5–10 kHz on S and T sounds.
- Thin: missing body between 60 and 250 Hz.
- Dull: missing presence and air above 2 kHz.
Every instrument spans several zones at once. A vocal has body around 200 Hz, its character in the mids, presence around 3 kHz and air above 10 kHz. EQ is deciding which zones each part gets to own. The zone-by-zone breakdown is in understanding the different frequency ranges, the most common problems have their own guides in the trick to fixing mud without losing warmth and how to fix harshness in your mix, and everything below 250 Hz is in the complete guide to mixing low end.
5Why should every EQ move be level matched?
Our ears think louder means better. That's a problem with EQ, because a boost can sound better simply because it's louder. Boost 3 kHz on your vocal and it sounds more exciting. Some of that is presence. Most of it is that the vocal got louder.
The fix is to compensate the output as you move, so the loudness stays put and only the tone changes. CARBON Q does this on every EQ move. As you drag a node it works out what the curve is doing to your loudness and compensates at the output, settling over a few seconds of audio and then holding still. Push 240 Hz up 4 dB and you hear a warmer channel. Pull it back and you hear it clear up. When a boost sounds better, it's because it is better, and when it doesn't you find out now instead of an hour later with fresh ears.
Level matching covers the EQ curve only. Each instance also has a Gain slider, and that one does exactly what you tell it. It's the control that sets your volume balance, so the two jobs stay separate: the curve changes tone, the Gain changes level.
The same principle runs through how to make good EQ decisions without expensive gear, which is worth reading if you don't fully trust your room.
6How do I EQ the whole mix at once instead of one channel at a time?
A level on its own doesn't tell you anything. A vocal at a certain level isn't right or wrong until you hear what it's sitting next to. The same goes for its EQ. So the most useful thing you can do is stop looking at one channel at a time.
Start by grouping your tracks. Every vocal through one vocal bus. Every instrument through one music bus. Every drum through one drum bus. All your subs and basses through one bass channel. Shaping those four against each other is where a mix comes together, because that's the level your listener hears.
CARBON Q is built around exactly that. It's one eleven band EQ spread across your whole session. A MIX instance sits on your output channel and hears everything together. Agents sit on your vocal, music, drum and bass buses and hear each part on its own. Neither is enough by itself, which is why they all talk to each other.
Every agent shows up as a live tile in the agent tray along the bottom of the window, with its own curve, preset bar and controls. The tray is in every instance, not just the MIX one, so you can work on the whole session from whichever window you've got open. You're looking at your mix curve, your vocal curve, your music curve, your drum curve and your bass curve side by side, at the same scale, while the song plays. If two channels are fighting, you can see both curves reaching for the same place.
Three controls make this practical to work in:
- Lock freezes a channel that's already where you want it. Its EQ becomes read-only and it stops taking preset changes, so you can keep working on the rest.
- Link, in the preset bar, ties an instance to the session preset. Switch it off to give one channel its own preset or reference track.
- Bypass on the MIX instance bypasses every instance at once, which is the fastest fair A/B you have of your whole mix.
Without a MIX instance an agent is guessing. It can hear your vocal, but not how that vocal sits in your song. So run one even when you're only working on a single channel. If you'd rather do this by hand with your own EQs, the same thinking is in reducing conflicting frequencies.
7How do I borrow the balance from a record I love?
We all have a record we hold our mixes up against. The vocal sits exactly where you want yours to sit. The kick and the bass share the low end without arguing. Getting your mix to sit the same way is the hard part, and it's the part CARBON Q was built to speed up.
Open the preset browser, go to the reference tracks tab, and drop the song in. From that one stereo file CARBON Q separates the vocals, music, drums and bass. It reads the tonal balance of each part and of the mix as a whole, and measures how loud each part sits against the rest. All of that becomes a preset. Select it and your session moves to match, EQ and level together. The stem separation and the analysis run on your machine, and your audio is never uploaded.
Use as much or as little of it as you want:
- The whole song. Let one reference set the volume and tonal balance across the mix, vocals, music, drums and bass together. It's the fastest way to hear how far your mix is from the record you have in mind.
- One channel. Want the vocal to sit like the vocal on your favorite song? Unlink the VOCALS agent and pick the reference there. The rest of the session carries on with what it had.
- Mix and match. Take the drums from one record and the bass from another.
- Audition. Load five references and flick between them until one of them makes the mix click, with every channel following each time.
A reference is a target to fast-track your foundation. It tells you where a record you trust sits. You're still driving the final sound. Choosing the right record matters a lot here, so read how to choose the right reference track for your mix and the complete guide to reference tracks.
8Where do good starting points come from?
Most preset systems have no context. A good vocal EQ can't be measured on its own. You only know a vocal sounds right when you hear it in the song it's sitting in.
So the CARBON Q targets come out of whole mixes, analyzed whole. For each one we capture both halves at once: the tonal shape of every element, and how loud that element sits against everything around it. The vocal, music, drum and bass targets that belong to a given record all come from the same mix, so they already work together. They aren't a pile of songs averaged into a single boring curve. They're reference points, and the ones you get are the ones closest to your audio.
Loop the busiest, most representative part of your song, usually the final chorus or the drop, and hit ANALYZE. A tray drops down where you choose which instances to analyze, and the button becomes GO. Press it, and CARBON Q compares your song against world class reference mixes, ranks everything it has, including your own reference tracks, and puts the ten closest Smart Presets in front of you.
Each one sets the tonal balance and the volume balance of every channel at once. The destination is fixed and the moves are worked out from your audio, so the same preset on two different songs gives two different sets of settings. Every preset is level matched as it lands, so stepping through ten of them compares ten tonal balances rather than ten volumes. Most of the time you know within four or five clicks.
9How do I use mid/side EQ across the whole mix?
A stereo file is two channels, left and right. Mid and side is the same information looked at a different way. The mid is everything both channels share, the center of your image: usually the kick, the bass, the lead vocal and the snare. The side is everything they don't share, the width around it: reverb tails, wide synths, doubles and room. Nothing gets added or thrown away. The signal is split, shaped, and put back together.
On a stereo EQ every move lands on everything. Add air and you brighten the vocal, the cymbals and the reverb together. Mid/side removes that compromise:
- Clear the low mids out of the sides and the mix stops feeling cluttered, while the center stays exactly the same.
- Open up the top of the sides for air without making the vocal harsh.
- Tighten the low end by pulling it out of the sides, so the sub and kick sit in the middle where they translate.
- Add weight or presence to the center without widening the reverb with it.
Mid and side has never been a difficult idea. It's always been slow, which is why most people try it once and go back to stereo. In CARBON Q both curves are on screen at once. The one you're editing is bright and the other sits dimmed behind it. Click the dimmed one and you're editing it. Both stay active the whole time. Right click or shift click the Stereo / Mid / Side selector and every instance in the session switches with it, so you get mid and side control across your whole mix from one window.
Run ANALYZE in Mid or Side and the Smart Presets give each half its own starting point. Drop in a reference and you get that record's mid and side targets, for the mix and for every role. Preset names carry a width tag as well as a tonal one, from Mono and Narrow through Focused Width and Balanced Width to Wide and Very Wide, so you know what a preset will do to your image before you click it.
Hear the difference
Listen to how adjusting the mid and the side can transform your mix.
Pop Mix · Mid/Side
Two cautions. Stereo is still the sensible default, and most of your moves belong there. And check in mono after any big side move, because side content cancels when a phone or a club system sums the mix. Boosting the sides lifts material that won't be there in mono at all. The full picture is in everything you need to know about mid/side processing and creative mid/side EQ techniques, and width itself is covered in the complete guide to stereo width.
10How do I hold the balance through the loud sections?
A normal EQ curve is set in one position, so it only handles the average of your track. It can't do anything about the chorus where the bass swells up over the vocal, or the snare hit that spikes at 3 kHz. For those you've always needed a different tool, and a different window.
When CARBON Q analyzes your audio it doesn't only learn the average tonal balance. It also measures where the sudden peaks happen. Turn on Dynamic Mode and it sets a threshold at those peak levels on every band. When your audio pushes a band past it, the band is pulled back into line in real time and released as soon as the peak has passed. The rest of the time nothing happens. Your tonal balance stops being an average and becomes something the mix holds through the loud bits.
The amount runs from 0 to 100%. At 50% it acts as a safety net, holding your mix to the tonal balance a preset or a reference put it at. Push toward 100% for tighter control. Use one amount across the whole range, or split it into low, mid and high, each with its own attack and release, so you can hold the low end firmly and leave the top alone. Dynamic has its own settings for Mid and Side too, so you can keep the width in check when the wide synths swell in the chorus without touching the center.
It only ever pulls down. We tried lifting bands that fell below their target as well, and it sounded unnatural, the way upward processing usually does. Pulling peaks back into line sounds like the mix was played that way. The wider idea of dynamic EQ is in fixing a mix using a dynamic EQ.
11What should I clean up before I balance?
Balance works best on clean sources. Two problems are worth fixing before you set the balance, because no amount of level or broad EQ will solve them.
Resonances
A resonance is a narrow frequency that rings louder than everything around it: a room mode in a vocal, a ringing snare, one boomy bass note. It only shows up on certain notes, so a static cut removes it even when it isn't ringing. RESO finds resonances and cuts them dynamically, only when they ring. The method is in how to find and fix unpleasant resonances.
Clashes that come and go
When two sounds only collide at certain moments, like a kick and bass landing together, a permanent cut thins one of them out for the rest of the song. FUSER listens to two signals, finds where they mask each other, and ducks only that band on the element you choose, only when the other one plays. It also works well for harshness that only flares up in loud sections. See how to fix channels fighting for space in a mix and how to use FUSER to soothe your mix.
Both go above CARBON Q in the insert chain. CARBON Q goes last on each channel, so when it analyzes it hears the cleaned-up signal.
12What's the workflow, start to finish?
About a minute to a strong starting point, then as long as you like making it yours.
- Group your tracks into four buses. Vocals, music, drums and bass.
- Fix the specific problems. RESO on anything that rings, FUSER on anything that clashes.
- Put CARBON Q last on each bus. Pick the matching role on each, then put one on your output channel and choose MIX.
- Loop the busiest part of the song. Hit ANALYZE in the MIX instance. The tray opens with the whole linked session armed, so press GO.
- Pick a starting point. Step through the ten Smart Presets, or drop in a reference track and select it.
- Take over. Nudge the levels, drag nodes, switch to Mid/Side, turn on Dynamic. It's only a starting point.
Why it goes last
When you press GO, CARBON Q listens to the audio coming into it and builds the balance from that. So it needs to hear whatever is already on each channel. Sometimes that's nothing, sometimes it's a whole chain of plugins. Put it last, on each bus and on your output, so the balance it builds is the balance you actually hear. If you add or change processing later, put it above CARBON Q and analyze again.
Transparent filters
Most EQs are minimum phase, which shifts the phase around every frequency you touch. On one channel you barely notice. Across five instances, one on each bus and one on the output, those small shifts stack up and the mix starts to sound a bit wishy-washy. CARBON Q uses our max quality linear phase filters, so boosting or cutting changes the tone without smearing your transients. Matched against a market leading linear phase EQ on a deliberately extreme +8 dB low shelf under 100 Hz, the two track within ±0.13 dB from 18 Hz to 2 kHz, and CARBON Q gets there at 41 ms of latency instead of 1,509 ms. The full test is published in MIXVAULT vs a leading EQ, which uses the same filters.
Common mistakes
- EQing to fix a level problem. If the vocal is too quiet, no presence boost will make it sit.
- EQing in solo. The channel sounds great alone and wrong in the song.
- Trusting louder. A boost that raises the level wins every unmatched comparison.
- Brightening everything. Four bright parts make a harsh record.
- Mixing without a target. Your ears adapt to whatever you've been listening to, and a dull mix starts to sound normal.
More of these in 7 EQ mistakes that can ruin your music, and what happens when they reach the outside world in why your mixes don't translate.
CARBON Q sets the volume balance and the EQ of your vocals, music, drums and bass from one window. Start from a Smart Preset built from your audio, or turn a record you love into a preset for your whole mix, then take over with every move level matched.
The walkthrough below covers every control, and the written version is the CARBON Q manual.
When the mix is balanced and you move on to the master, the same ideas carry over. The complete guide to audio mastering covers that stage, and if you're working from stems, how to stem master a song shows CARBON Q across a set of finished stems.
?Volume balance and EQ FAQ
What is volume balance in a mix?
Volume balance is how loud each part of the mix sits against the others: the vocal against the music, the kick against the bass, the drums against everything else. It's the first thing a listener hears, before tone or effects, which is why a simple mix with a great balance sounds more professional than a heavily processed one with a poor balance.
Should I set the volume balance or the EQ first?
Volume balance first, most of the time. Changing a channel's level changes how its tone reads, so if the levels are wrong you'll end up EQing to compensate for something a fader would have fixed. Once the balance is close, the EQ moves you need are smaller and easier to judge.
Why does my EQ sound great in solo but wrong in the mix?
Because tone is relative. A channel only sounds bright, heavy or thin compared to what's around it, so an EQ that sounds perfect in solo often fights the rest of the mix. Use solo to find a problem, then make the tonal decision with the whole mix playing.
What is level matching and why does it matter for EQ?
Level matching compensates the output of an EQ so the overall loudness stays the same as you move the curve. Our ears hear louder as better, so without it a boost can sound like an improvement when it only made the channel louder. With it, the only thing that changes is the tone.
Does CARBON Q move my DAW faders?
No. Each instance sets its own output gain on the channel it's sitting on, so your faders and automation stay where you left them. Bypass any instance, or the whole session from the MIX instance, and you're straight back to what you had.
If EQ moves are level matched, how does CARBON Q change my levels?
They're two different controls. Level matching cancels the loudness an EQ curve adds or removes. The volume balance lives on each instance's Gain slider, which is what a preset or a reference track sets and which level matching never touches.
Can I use a reference track for just one channel?
Yes. Switch Link off on that agent, open the preset browser there and pick the reference. The rest of the session carries on with what it was using. Do the same on two agents with two different references if you want the drums from one record and the bass from another.
Do I need a CARBON Q instance on every channel?
No. Use the roles that match your session. What you should always have is the MIX instance on your output channel, because it calculates the Smart Presets from the whole session and shares them out. Without it, each agent is working from its own channel alone.
Will mid/side EQ cause problems in mono?
In mono the side content cancels, so cutting the sides changes nothing when a phone or club system sums the mix, while boosting them lifts material that won't be there in mono at all. Check your mix in mono after any big side move.
Is Dynamic Mode a compressor?
It compresses, but only where and when a band goes over its target, so it behaves more like an EQ that holds its shape than a compressor across the whole channel. It never lifts anything. If the audio stays under the target, it does nothing.
Does my audio get uploaded anywhere?
No. Everything in CARBON Q runs on your machine, including the reference track analysis and the stem separation. Nothing leaves your computer.
Balance first, then tone
Balance gets skipped because it feels too simple to be the thing that's wrong. Usually it's exactly what's wrong. Check the levels before you touch another EQ. Then shape the tone of every channel against the rest of the mix and against a record you trust, with every move level matched.
Do that across your vocals, music, drums and bass together and the mix stops being a pile of channels you EQ one at a time. It becomes one picture where every part has its place, which is what every record you love already sounds like.