The proven system thousands of producers and engineers use to create release-ready mixes from anywhere. No expensive monitors. No treated room. Just a workflow that translates.
When you pair proper calibration with disciplined referencing. Headphones remove room issues but exaggerate left and right separation, so your goal is to learn how that translates to real-world playback. Once you do, it's a superpower.
Four moments you'll actually run into. Tap the one you're in.
Reach for the
Quick wins: work quieter than you think. Use MIXRef to pick references that serve your client's ears and your own, and compare 8-12s loops against them. On marathon sessions, switch to non-critical work like editing and let your ears reset.
Calibration evens out the peaks and dips baked into your headphone model, so you mix from a more consistent starting point. Unit variation, pad wear, how well they seal and your own anatomy still move the result, so treat this as a better starting point, not a promise of neutrality.
Watch the white line. That's what reaches your ears. Example curve only: every model measures differently, so look up your own headphones.
Flat sounds like the honest answer: change nothing, add nothing, hear the truth. It's a trick question. Flat where? Your mix passes through a room, a head and an outer ear before anyone hears it, and no ear on earth receives it flat.
Think about a great record on great speakers in a great room. The room lifts the low end. Your head, shoulders and outer ear push the presence region up by 8 to 12 dB, depending on whose head you measure.
Every record you've ever loved arrived at your eardrum wearing that shape. That shape is what your brain files as correct.
A headphone tuned flat at the eardrum strips all of it back out. That's not neutral. It's a sound that has never once reached your ears from a loudspeaker, and it reads exactly the way you'd expect: thin, harsh and strangely lifeless.
The fix came out of a seven-year blind listening program at Harman's research lab in Northridge, California, led by Sean Olive, Todd Welti and Elisabeth McMullin.
| 2009 | The old lab standard cracks: 80 listeners prefer the diffuse-field target's wide 12 dB presence peak cut to 3 dB. The hunt for a better curve begins. |
| 2012 | The program begins. Measuring hundreds of headphones over the following years shows price predicts almost nothing about frequency response. |
| 2013 | First blind result: listeners prefer an accurate loudspeaker in a good room, measured at the eardrum, over the free-field and diffuse-field lab standards and over high-priced headphones as shipped. |
| 2014 | The test goes wide: 283 listeners in Canada, the United States, Germany and China, every brand hidden behind one calibrated test headphone. |
| 2015 | Listeners of every age and experience level dial bass and treble to taste, program after program. The curve absorbs what their ears keep asking for. |
| 2018 | The validation: 31 headphone models, 130 listeners. In 28 of the 31, listeners prefer the version EQ'd to the target. The ties were models already near the curve. |
This is the measured target, not an artist's impression of one. Pick a region to read it.
Four of David's released mixes, loudest 18 seconds of each. Press play, then flip.
It's an average, not your ears. Your pinna and canal are yours alone, and above roughly 8 kHz they dominate. That's why the chart fades there and why no preset lands the top octave perfectly.
It's married to its rig. The curve means something only on the GRAS rig that defined it. Newer measurement heads hear differently, and Harman derives fresh targets for those rather than reusing this one.
Preference isn't a law. A third of listeners wanted more or less bass, the early panels leaned heavily on Harman's own staff, and Olive himself calls the curve a guideline rather than the last word.
| DF tilt | The diffuse-field standard, modernized with a downward tilt. The measurement-purist school on newer rigs, and Harman's own newer-rig pilot work landed near it. |
| Studio | Sonarworks SoundID, Slate VSX, Realphones chase monitoring translation, not listener preference. If your SoundID profile disagrees with an Oratory1990 preset, this is why. Both are legitimate; pick one and learn it. |
| Your ears | Individual calibration. Above a couple of kHz, a population average stops being you, so this school measures or tunes each listener. |
| The 36% | Harman's own dissenters: the more-bass and less-bass camps in the split above. They're data too. |
So calibrate to it, then stop worshipping it. The target removes your headphones' worst lies and hands you the measured stand-in for good speakers in a good room. Your references settle the rest.
Four steps, in order.
Look up your own headphones on the Oratory1990 preset list. Hundreds of models are covered, Sennheiser through Sony.
Turn it into something your plugin can load. Most EQs take Oratory's numbers exactly as printed. FabFilter Pro-Q is the exception: its Q is scaled differently, so typing the printed values straight in gives you a curve that isn't the one you looked up. The converter below does that maths for you.
Headphone output only, never the mix bus. Put it upstream of your loudness meters and they start reading your monitoring path, not your export, so a true-peak you thought you checked never actually was.
Toggle bypass and match levels before evaluating. Louder isn't better, just different.
Frequency and gain you type in exactly as printed. Only Q changes, and only because Pro-Q measures it differently. Type the number from your preset.
Bell filters. Pro-Q's bell Q is Oratory's multiplied by 1.4142, which is the square root of two. Shelving filters convert differently and are not in here yet.
Three buses, climbed in order. Mix A holds the mix together, Mix B supplies the slam, and the master is the last chance to affect the sound.
Why this comes before referencing. Most mixes arrive at the two bus still too dynamic, without enough grip at track or bus level. That's a mixing correction, not a mastering trick. Until it's fixed, a quiet mix against an industry master is apples to oranges.
A stereo aux, not the Master Fader. Every track feeds it. This is where the mix stops being a stack of parts and starts behaving like a record.
Hit it pretty hard. Red-line it, take it to extremes so you can actually hear what it's doing, then back it off. Usually a little more than slightly.
The heart of the two bus, and the one worth dialing in properly. The full method, on a real mix you can hear, is in the Bus Compressor Trainer.
After the compressor, so it's working on a signal that's already behaving.
Check the phase modes while you're in here. Listen for whether you can hear a difference in quality between modes on the two bus, versus what the same modes do at track level. They aren't the same decision.
Some songs want it. Some don't. This one is a taste call, not a step.
Dialed in over on the referencing plugin, then moved onto Mix A to live there permanently.
Two dB and calling it done teaches you nothing. Drive it hot enough to hear, then retreat on purpose.
Six steps, one compressor, and audio you can actually hear rather than a diagram of it. It is coming back here, so check in on it.
Drag anything. The picture follows.
A worked diagram, not a recording. The gain-computer maths are real.
Set it and leave it. Everything else gets found one variable at a time.
Stays put for the rest of the dial-in.
Push past subtle, on purpose. A needle that barely moves teaches you nothing.
The zone worth living in while you listen.
Park it. One less variable while you dial the attack.
Slowest to fastest. Hunt where the grab fits the song instead of fighting it.
Usually the slowest setting, or the one right after.
Auto was a placeholder. Pick a lane.
Exaggerate until you can hear it. Retreat until only the good part is left. You cannot find a line you have never crossed.
Some songs want the grip, Andy Wallace SSL territory, where it becomes the tone, not just control.
Mix knob: usually left at 100%, wet.
Five real sounds, four compressor families. Drag anything.
The kick, snare, bass and groove are mine, straight off my rig and uncompressed. The vocal is not mine: it is a live performance someone else captured on a live mic, lightly processed on the way in. Nothing here has been compressed, which is the part that matters, because it means you are hearing this compressor do the whole job rather than finish what a tracking chain already started. One sound at a time on repeat, because a full mix hides the mechanism: every hit overlaps the last one's decay. The picture shows the compressor's raw output against the untouched hit, and your ears get the two level matched so the A/B is a fair fight. The compressor is real, but it is an archetype, not a model of any specific unit. It teaches the behaviour. Your compressor supplies the colour.
Feed-forward and predictable: the attack and release knobs do exactly what they say. The bus-glue archetype, like an SSL-style bus compressor.
Nothing to look at now. Press play, turn a knob, and stop when the sound is right rather than when the number is. The meter and the waveform will tell you what you did the moment you switch this back off, and that gap between what you heard and what you see is worth more than any reading you could have taken while watching.
Nothing is compressing yet. Turn the threshold down.
This stage runs on whichever family is selected, so you can hear what saturation on its own does to a clean VCA long before you reach for Annihilation. Two blend knobs on purpose: the saturation has its own mix and the compressor keeps its own. And the order is a sound. Saturate first and the compressor gets hit with a denser signal. Try both ways.
Attack and release only mean something together. These are the four corners. Press one and watch the lower curve.
It filters what the detector listens to, not what you hear.
| High-pass point | When |
|---|---|
| 30 to 40 Hz | Sometimes enough, when the low end is already behaving. |
| 70 to 100 Hz | Where it usually ends up. |
| 120 Hz | When the kick is really driving everything. |
I mix low end heavy on the kick, so the compressor hears the low end first and makes most of its decisions on it. Left alone, the kick ends up driving the whole two bus and the snare barely gets a vote. Raise the filter and the kick stops dominating the decision, so the compressor delivers an even amount between the kick and the snare. That matters because kick and snare drive most of my mixes. On a vocal forward song, that weight shifts to the vocal and you dial it accordingly.
The test is visual. Watch the needle while the song plays. You want to see it move with the kick and with the snare. If it's only twitching on the kick, keep raising the filter until the snare starts pulling it too.
Each one sets the rack above and then tells you what it did. The numbers aren't the lesson. The lesson is that the same few decibels of gain reduction can arrive four different ways, and every one of them sounds like something.
The module above teaches one sound at a time. This one is a whole record: my own mix of New Revival printed with nothing on the two bus. Hear three settings I actually used, including Slam and Settle, then recreate them yourself by ear on the same song.
Open itA parallel copy of the whole mix, compressed far past anything you'd do in series, then blended back underneath.
| The move | The detail |
|---|---|
| The routing is the trick | Mix A feeds Mix B pre-fader, untouched by Mix A's fader, so you can mute Mix A entirely and still hear the slam alone at full volume. |
| What goes on itan 1176-style or your fast FET compressor | The same bus compressor again, or an 1176-style one. Full signal, no mix knob. Often stacked: a slow attack, then a medium behind it, gripping in stages.
 |
| It moves across the song | Back it off on the intimate sections. Push it at the last chorus if the song needs a lift, sometimes with width or volume alongside, so there's another level to go to. |
The question you're answering isn't "is this too much." It's "how can I slam this mix and still make it sound good?"
This is what the pre-fader send bought you. Judge the slam on its own terms.
Not acceptable. Great. If it sounds bad alone it won't become useful blended.
Back to the real mix, with the slam parked at silence.
Usually not much. It depends on the song and how loud you're going.
Everything above lands here. K-Clip into a Pro-L is on from the first bar, and I mix through it.
On from the first bar. Shave the peaks before the limiter ever sees them. Sometimes a little, sometimes a lot, and a great deal of the density comes from here.
An evaluation, not a tone pass. I let the assistant scan against the Jesus Is My Help target, then keep only what helps. The EQ goes off. The imager almost always does, it pulls the mix in. The exciter stays, dialed.
Its own instance, never the settings the scan handed it. IRC5, driven hard. If the limiter is doing all the work, everything above it was too polite.
Character is the one slot on that rack that's a taste call rather than a step. Four flavors of it side by side, with the harmonics each one actually adds measured off the audio instead of drawn from memory.
Open itThen back it off until it doesn't. Ears only, no meters.
That number picks the next move: slam the parallel harder, take low end out and push, or back off and keep what already sounds good.
Back to the Pro-L, same process. Then the AOM Invisible. The method travels, the plugin isn't the point.
Want to see it running? The walkthrough further down takes a real mix up this ladder, on camera.
Integrated LUFS across the whole song, so you know the ladder is working.
Where mixes land before gripping starts. Some genres stay here on purpose.
Where most land once gripping starts. Respectable, and it holds.
Where I want you, without a fight. Not every song has to sit here.
Reachable with low end work and harder gripping. Reachable isn't right for every song.
I don't love sitting at -4 or -5. I'm a fan of dynamics and equally a fan of glue and impact, and those two argue constantly. The song, the artist, the vision and what the instruments are giving us decide. Not every song should go that hot: plenty sound terrible that loud and want a dB or a handful more open.
Aim at your end goal, not at mine. Decide honestly whose loudness this is: yours, the client's, or what A and R told them they need. All three are legitimate. Make the call with your eyes open instead of defaulting to whatever your ears wanted that day.
No console, no soffit-mounted mains, no treated live room. A desk, a screen, a pair of open-backs and the discipline above. Everything the ladder asks for happens here.
Two different jobs share the word loudness, and running them together is why level matching goes wrong for so many people.
| The job | What it actually is |
|---|---|
| Building your loudness | Everything above. Saturation, bus compression, a parallel slam, clipping into a limiter. It's a tone decision, and it changes the mix on purpose. |
| Matching the level | A measurement decision. Clean gain only, or pull the reference down to meet you. Nothing that changes tone. |
The moment you reach for saturation, parallel compression or a limiter to close the gap, you've changed the thing you were measuring. It stops being a controlled comparison. Build first. Match second. Then judge.
Skip the match altogether and you'll pick the louder one every single time, and call it taste.
MIXRef Pro hands the A/B over already level-matched, so you barely touch this fader.
Your ear isn't a meter: how much low end and top end you hear changes with playback level, which is why one mix feels like two mixes at two volumes. Drag the level and watch what your ear quietly stops hearing.
So what? Loudness is spectral, not just level. 2 to 5 kHz reads louder than the same number sitting in the sub. Quiet checks work the same way: only the mids survive.
FFR™ takes A/B referencing further. Instead of comparing entire mixes, you take a magnifying glass to one frequency zone at a time and reverse engineer what's happening against a mastered release.
Zooming in by frequency only helps if you know what your EQ is doing when you get there. One filter, three archetypes: a clean parametric, a proportional-Q that narrows as you push it, and the Pultec trick of boosting and cutting the same spot. The curve is computed from the real coefficients, not painted on.
Open itGet the timing backwards and referencing lies to you: too late and you're not influenced, too early and you can't trust what you hear.
| When | What it's doing there |
|---|---|
| From the first minute to the lastVision and direction | Shapes the picture in your head: arrangement, drum tone and sound, finding samples, establishing guitars. Useful at the very beginning of a mix and still useful at the end. Reference freely, and often. |
| Only once the two bus is workingLevels, balance and sonics | These only become trustworthy once your mix is already at industry loudness, which is what The Loudness Ladder is for. Compare a quiet mix to a mastered release and your magnifying glass is going to lie to you. |
So the running order is simple. Reference early for what you can learn from it. Once the rough is built and you're bobbing your head, close to printing mix one, do the full two bus and get it loud. Only then bring the references back for levels, balances and sonics, and take the magnifying glass to them with FFR™.
Start with a handful. Four or five. Ten isn't too many while you're still discovering what the song wants to be, and flipping between them is being inspired, not indecisive.
Then narrow. By the time you're matching levels you want the two or three that genuinely match this song, not the ten that inspired it. A song being in G tells you almost nothing about how much low end is in the record, or where the voice sits inside it.
| The criteria | Which means |
|---|---|
| Match on tessitura | Where the vocal actually sits and lives, not the key written on the chart. This is the one people get wrong. |
| Match on arrangement | What's actually playing, and how much of it. |
| Match on drum tone | The sound of the kit itself, not just the groove. |
| Male against female vocal | Different range, different weight, different everything downstream of it. |
Everything above is the manual version, worth doing by hand once because the judgement is the skill. Here's the same job done against the measured session: MIXRef Pro ranks real records against what it actually hears, tells you why each one made the list, then labels what it measured from your audio against what it only inferred, so you always know which half you're trusting.
The same job the section above just walked through, done against the measured session rather than from memory. Note the labels: Jireh is where to calibrate first, Great Things is in the list as the low-end anchor, Promises for its louder ceiling. Each reference is there for a stated reason, which is the difference between a reference set and a playlist.
It also hands off already level-matched, the part that decides whether the comparison is worth anything. The reference gets pulled down so the two are fair, and FFR™ is one button away.
A reference isn't an answer, it's the beginning of a question. When somebody hands you one, ask what specifically they like about it. The drums? The weight of the vocal? The polish? People mean one narrow thing by a reference. Give them the whole song back and they'll tell you it's wrong, and neither of you'll know why.
Here's the one I run into constantly. I get handed CCM references, Bethel, Elevation, The Belonging Co, for a song that's actually black gospel. Give them exactly what they asked for and nine times out of ten they come back and say no, I want the gospel snare, I want the gospel kick drum.
The stated reference and the intended sound were two different things the whole time. Nobody was being difficult. They reached for the records they knew how to name.
Ideally the references you're given are genuinely in the ballpark of the song. When they aren't, catch it before you mix to the wrong target. Read between the lines, and ask better questions up front.
Load Pro-Q, or any EQ with steep slopes, after your limiter. Use high-pass and low-pass filters at 48 dB per octave or higher to isolate zones, then A/B between your mix and the reference. One range at a time.
This is that exact move on a real mix: same track (Floor 1.cm), same Pro-Q 4, five passes. Isolate the top, sweep it into place, bracket a band with both filters, widen it, then flip to isolating from the low side. Watch the slope readout: that steepness does the isolating, not your imagination.
1 / 5
A High Cut alone, 72 dB/oct, positioned low (around 30 Hz) on Floor 1.cm. One filter, nothing else. This is what the tool looks like before you start bracketing a zone.
Same High Cut, still 72 dB/oct, swept up to sit right around 100 Hz. Moving one filter into place before you add the second lets you hear exactly where the wall lands.
The core FFR move: a Low Cut around 100 Hz and a 72 dB/oct High Cut around 1 kHz, boxing in one band. Everything outside the bracket drops away, so you can A/B just that slice against the reference.
A wider window at a gentler slope. Low Cut around 500 Hz, High Cut confirmed at 5988.8 Hz (Q 1.000, 48 dB/oct). Good for checking a broader range, like the whole midrange, in one pass.
The bracket comes apart. Low Cut only, 30 dB/oct, positioned around 2 kHz, isolating everything above it: the mirror of where step 1 started.
The five FFR™ zones, one at a time, on a real mix. Pick a zone and everything outside it drops away, which is exactly what your filters will do in the DAW.
One honest thing about these five. They're diagnostic spot-checks, not full coverage. There are real gaps: 70 to 100 Hz, 250 to 400 Hz, 800 Hz to 2 kHz, 5 to 8 kHz and everything above 12 kHz sit in none of the brackets. Getting all five to match doesn't prove your whole spectrum matches. These are simply the places problems hide most often.
This is where the kick's weight lives, sub-bass included. Too much here and the mix feels bloated instead of powerful. Too little and the kick disappears the second it plays through a real speaker.
Warmth lives here, and so does mud. This is the one range that trips up almost every home mix. Push it for more fullness and the whole song turns to soup.
Presence without the boxy, cupped-hands sound. Too much here and vocals and guitars sound like they're singing into a cardboard tube.
Vocal bite and snare snap live here. Too much and the mix turns harsh, tiring to listen to after a couple minutes. Too little and nothing cuts through a car stereo.
The open shimmer on top of the mix. Too much and it turns to hiss. Too little and the whole song sounds like it's playing through a blanket.
Same five zones, no audio required. Point at a range on the axis, or tap its name below, and read what it does before you ever open a filter.
Work one range at a time. Once low end, mids and top end feel balanced like the reference, remove the filters and re-listen full-range. You'll instantly hear how your mix translates.
Load the NS10 preset, then the Avantone, then something like VSX or another modeling solution, and run the same range through each for a different perspective. Check the reference through them as well, and take notes as you go. It's powerful.
FFR™ hands you the magnifying glass. The Critical Listening Survey tells you what to look at through it.
This is the document I use to pull a reference apart, and it's the step almost everybody skips. Fifty nine questions in my own words: which instruments actually play in the intro versus the final chorus, where the kick fundamental sits, how much bottom end sustain it has, what frequency the snare crack comes from, room or plate, how wide the overheads are panned, where the lead vocal lives.
The point: stop saying "it sounds bigger" and start writing down numbers you can go and look for in your own mix. MixPrint measures that same record's balance for you. The survey teaches you to hear what it found.
Four of the survey's questions, wired to the frequency map above. Drag each slider to where you hear that element in your reference, and it plots against the five FFR™ zones. That's the survey turned into a picture.
After balancing each range, test your mix on multiple systems. The goal is translation, not imitation. When it feels balanced and emotionally consistent everywhere, you've done your job.
HeadphonesCalibrated. Your first truth.
The carWhere low end goes to confess.
EarbudsHow most people will hear it.
MonitorsThe room's second opinion.A real mix, on camera: calibration, Frequency Focused Referencing™, and the whole climb up the ladder. Mix A, Mix B, the mix bus and the main outs.
One honest note. This was recorded a while back, so a few of the plugin picks on screen have moved on since. The routing hasn't, and neither has the method. You're watching the same architecture this guide just walked you through. The plugins change every couple of years, the ladder doesn't.
Three quick missions to make translation second nature.
Pick 5-10 great masters across your genres. Loudness-match them and keep them ready in every session.
Loop 8-12 seconds, level-match, compare by frequency zone, make a single decisive move, re-check.
Bounce a pass, listen on 2-3 real-world systems, note the deltas, and iterate once. Then get feedback on it in the Control Room.
You already opted in, so there are no hoops: everything downloads now, and this page stays free to come back to.
Calibration plus FFR™ is the exact system David uses on client records and teaches live every week. Featured on The Pro Audio Files, Produce Like A Pro, and URM Academy.
Everything you just practiced by hand, set up for you. Keep scrolling, the tools take turns.
Drop a folder of multitracks. Naming, organization and gain, automatic, before you hear a note.
Click to see it properly
New RevivalSource WorshipCALIBRATE HERE FIRST
You AreSource Worship-11.3 LUFS
UfanelweDr. Tumi-10.4 LUFSYour references, measured and ranked for the session in front of you. Calibrate first, then mix.
Click to see it properlyReads the session and triages it. What needs a repair, what's your call, and what's handled for you.
Click to see it properlyEvery note timestamped and pinned to the waveform. The phone companion is on the way.
Click to see it properlyYou don't need another course. You need a room of mixers who critique and grow with you every week.
Three ways in
MIXPrepThe Mac app that gets sessions ready to mix.
Not for you if your problem is the mix itself. This fixes prep, not decisions.
The Control RoomEverything I teach, plus MIXPrep.
Not for you if you want it done for you. This is the craft, and it takes work.
Multi-Platinum Mixing and MasteringMixed and mastered by me.
Or I mix it with you, on camera, teaching every move: Mixing Discipleship.
Not for you if the recording isn't there yet. I'll tell you straight.
Between them: the Mix Mastery Blueprint, one to one on your own mixes.
Asked to name one mixing channel to follow, Doug Weier named David.
Doug Weier 3x Grammy nominated · 21 Billboard #1s · Chris Tomlin, MercyMe, Tauren Wells, Walker Hayes
“It was night and day different. Night and day. It was huge.”
Lucas Produce Perform, worship music production · after a Mixing Discipleship session · one-on-one, on your own mix
“I've spent time at Full Sail, worked alongside pro engineers, and logged hours in multi-million-dollar studios. But my biggest wins came from watching David mix. Something about his workflow just made it all click.”
Tom Camp Producer, Engineer, Musician
“After David's tutorials, I started getting paid clients. I did 2 years of audio school, and I learned more from ONE of David's courses than all 6 years combined.”
