Mixing on headphones: crossfeed, correction software and the limits nobody fixes

Plenty of finished records now start life in a pair of headphones: on a train, in a rented room, at 2am when the neighbours are asleep. Headphones are cheap compared with treating a room, they travel, and they show you detail that a bad room hides. They also lie in a few specific, predictable ways. This guide covers what those are, what the main families of software fixes really do, and where no plugin can help you.

If your problem is the room rather than the headphones, our guide to budget acoustic treatment is the better starting point. If you are still learning what to listen for, the bedroom mix mistakes guide pairs well with this one.

Why headphones sound different from speakers

With a pair of speakers, each ear hears both speakers. The left speaker reaches your right ear too, a fraction later and duller, because your head is in the way. Records are balanced on that kind of system, so that blend is baked into how a mix is judged.

Headphones remove it. The bs2b project, an open-source crossfeed filter, puts it plainly: stereo recordings are mixed for loudspeakers, and on headphones what is missing is "the sound going from each channel to the opposite ear", which would normally arrive slightly later and with some high-frequency roll-off. The same page argues that long headphone sessions get tiring more because of this "superstereo" effect than because of poor headphone design.

In practice that produces a few familiar habits:

• Panning too wide, or too timid. Hard-panned parts sit inside one ear on headphones, so you either pull them in more than speakers would need, or you never notice how far apart they are.
• A centre image inside your head. The lead vocal and kick sit between your ears rather than in front of you, which makes level and depth calls harder.
• Reverb and low end read differently. With no room adding its own reflections and bass build-up, it is easy to add too much reverb, or to misjudge how much sub a speaker system will actually reproduce.
• Fatigue. Isolation plus volume is a recipe for ears that stop telling you the truth after an hour.

None of this means headphone mixes are worse. It means the reference is different, and you have to correct for it on purpose.

Fix one: crossfeed

Crossfeed is the oldest and simplest fix. It feeds a filtered, slightly delayed copy of each channel into the opposite ear, imitating what your head does with speakers. bs2b describes its own version as low-pass and high-boost filters that send part of each channel to the other ear with a small low-frequency delay, so the sound seems slightly outside the head. It notes that it does not add ambience or full head-related (HRTF) processing.

Commercial plugins give you more control. Goodhertz's CanOpener Studio is a crossfeed processor modelled on an idealised loudspeaker and room, and you choose how much of the effect to use. Goodhertz suggests starting at a 60° angle with 100% crossfeed, and notes that a 0° angle gives the flattest frequency response. It also bundles monitoring tools such as mono, left/right flip and polarity switches.

How to use it: put crossfeed on your monitor path (a master-bus insert you bypass before export, or your interface's monitor mix), never on the mix itself. Pick a setting, and leave it there for the whole project so your ears adapt to one reference.

Fix two: headphone correction curves

Every headphone model has its own frequency response, with peaks and dips that colour what you hear. Correction software applies an EQ curve that aims to flatten that colouring toward a target.

Sonarworks' SoundID Reference is the best-known example. Its product page lists support for more than 500 headphone models, and says calibration works by applying "a pre-made calibration profile" in the software. Sonarworks also sells individually calibrated headphones as a separate option. The software runs as a DAW plugin or as a system-wide app, lets you adjust the target curve, and offers zero-latency, mixed and linear-phase filter modes. It also includes translation checks that simulate other playback systems such as car stereos.

Two things to keep in mind:

• A model profile is an average. Units of the same model can differ, and pads wear and fit changes how a headphone sounds on your head. A profile gets you closer, not to perfect.
• Phase modes cost something. Linear-phase filters add latency, which matters while tracking. Use the low-latency mode when recording and switch when mixing.

Fix three: virtual rooms and head tracking

The most ambitious tools simulate an entire speaker setup in a room. Waves describes Nx as a "virtual acoustic mix room for mixing on headphones". It can track head motion so that the sound changes as you turn, using your computer's camera or the company's Nx Head Tracker, and it lets you enter personal head measurements to adjust the 3D effect to your anatomy.

Head tracking matters because a fixed simulation still moves with your head, which is one of the cues that tells your brain the sound is inside it. Whether a given virtual room helps you is personal: try any trial version on a mix you know well before trusting it on a new one.

What none of these fix

Even with all three layers stacked, some things stay hard on headphones.

• Physical low end. You feel sub-bass on big speakers and in clubs in a way you do not on headphones. Check the bottom octave on a meter or spectrum analyser, and on a second system.
• Your own hearing. Software corrects the headphone, not your ears. The WHO's fact sheet on hearing loss says over 1 billion young adults are at risk of permanent, avoidable hearing loss from unsafe listening. The WHO-ITU safe listening standard it promotes describes a "sound allowance" feature that tracks how much of a reference exposure you have used. If your phone or OS has such a meter, turn it on.
• Translation guarantees. No simulation is a real car, a real phone speaker or a real club. It is a better guess, not proof.

A working routine

1. Set a level and keep it. Mix at a moderate, consistent volume, and take short breaks every hour or so; if a mix sounds dull after a break, you were probably listening too loud.
2. Monitor path only. Correction, crossfeed or a virtual room go on the monitor output and are bypassed for the bounce. Write it on a sticky note if you have to.
3. Use reference tracks. Load two or three released songs in your genre, level-matched, and flip between them and your mix through the same chain.
4. Check mono often. Mono collapse exposes phase problems that headphones can hide. Our mid/side guide covers how to look at the side channel on its own.
5. Second-system check before you sign off. Phone speaker, laptop, car, cheap earbuds, a friend's speakers. Note what changes, adjust, and repeat.

What we think

Headphone mixing is not the poor relation of a "real" studio. For a lot of artists, especially outside the few cities with affordable room hire, it is the studio. The tools above are good, and we think a modest crossfeed plus a correction curve is worth trying for anyone mixing mostly on headphones.

But the software sells best when it promises to replace the room, and that is the part to be sceptical of. The most valuable habits in this guide are free: consistent level, reference tracks, mono checks and a second system. A plugin that lets you skip those is not helping you. Spend the first money on comfortable, closed or open headphones you will actually wear for hours, and on protecting your hearing, because no correction curve can give that back. Then add software one layer at a time, and keep only what makes your mixes translate better.