Hi everyone, I'm bored tonight, so I'll share what I know about mixing with headphones and IEMs. As many people don't have the luxury to work in decent rooms with calibrated speakers, headphones are often to preferred listening system for most beginners, and people like me whose mixing room are too small to have a correct bass response. I hope this will be useful to you guys, as this is the knowledge I could've used ten years ago, and it would have prevented me to waste thousands on the shittiest pairs of headphones. Having a listening system that you can trust is one of, if not the most important thing in working with sound.

1) What are the practical differences between listening on speakers and headphones/IEMs ?

• Completely dry sound. Because of the lack of room acoustics. You will tend to put more reverb than on most music made on speakers, as you will try to compensate for the lack of space. Also, fast and important changes in volumes will be fatiguing. A good example of this is is the drop in Watch Out For This by Major Lazer. Absolute ear rape on headphones, but it probably sounded okay in the studio, as room reverbation smoothes dynamics.

• Binaural imaging. Headphones have the largest stereo image possible as there's no crossfeed between the left and right channels. Some people will be more reserved when it comes to panning while listening on a binaural system. For example, while LCR panning is good mixing technique for speakers, it sounds a bit too extreme on headphones in my opinion

• Lack of tactility. I saw Glass Animals live in 2016. While my ears were feeling okay, I thought my ribs were gonna break due to the sheer amount of sub-bass thrown at me. That feeling is only possible on speakers as headphones and IEMs don't direct sound to your whole body. It makes it hard to judge sub-bass volume and transients without speakers.

2) Considering those differences, how should we mix on headphones/IEMs ? (And why speakers emulations plugins are dogshit)

The first thing to have in mind is that a good part of the general public is listening to their music on headphones and earbuds. There's no statistic to my knowledge, but in 2023, it's safe to assume that it's even the majority (if we exclude fuckers listening on their laptop/phone speakers). Because of that, we should not see headphones and IEMs as second grade listening systems. The ideal listening setup for mixing and mastering should therefore feature both speakers and headphones/IEMs.

But what if you don't have the room and the speakers ? Well I'm sorry, but there's no way for headphones and IEMs to replicate the sound of speakers in any realistic way. All tools like Waves NX, Slate VSX, Acustica Sienna etc... are in no way close to the sound of speakers. Let me explain :

​

These are overlaid HRTFs from different subjects at seven different angles. We see that the variations are immense in all angles. That means the sound arriving at the eardrum is extremely different from one listener to the other. That's one of the biggest problems with plugins that try to emulate the experience of listening to speakers on headphones. They are based on average HRTF curves that can hugely differentiate from yours. It will immitate the experience, but never truly feel like you're in a room listening to speakers in front of you.
That means that the imaging won't be right either. A stereo speaker system in a decent room has the ability of creating a phantom image in the middle when a sound is completely in mono. Try that on those plugins, it doesn't work. Your brain only knows your own HRTF curves, not the ones they used when creating those plugins.
Also, due to the lack of tactility, it also won't replicate the feeling of sub-bass going through your body (and don't even talk to me about things like Subpac).

In conclusion, when mixing only on headphones and IEMs, make the mix sound good on them. Just don't be stupid and keep in mind that your music can be listened in big rooms so don't drown your whole mix with reverb if it's not intended. The only speakers "emulation" I personally acknowledge in any meaningful amount is the Phonitor Matrix system from SPL, that only creates crossfeed and delay between the left and right channel. It's easy to emulate with basic plugins, and it's useful to check if your mix is wide enough for stereo speakers.

3) What should we look for in good headphones/IEMs when it comes to mixing ?

The biggest complaint we hear about mixing on headphones is that they don't translate very well on other supports, even other pairs of headphones. Let me explain clearly why in very simple terms :

Most headphones are absolute horseshit.
​

There's a way you can objectively know if your headphones are good or not. It's with measurements. With the graphs of frequency response, expanded phase response/group delay, THD and CSD (waterfall), you can accurately track the sound quality of a device, and boy does it look bad on most headphones (there are other types of measurements but these will tell most of the massacre). I mean look at the measurements of the DT770, one of the most popular pro headphones. It's dogwater. Or look at the ones of the Focal Clear Pro Mg, a 1000€+ pair of headphones destined for critical listening. It's so bad it could make people deaf. It really shows that expensive equipment is not always better. Don't know how to read measurements ? Let me explain them by looking at the graphs of the HD650, a known "reference" headphone :

Let's start with the simplest graph to read, THD :

​

THD stands for Total Harmonic Distortion. When you want to hear a 50Hz sine wave, you want to hear that and nothing else. THD indicates how much harmonics the headphone produces at any given frequency. It also tells at which level your headphones will start to compress certain frequencies (when it can't get louder, it distorts). So the goal is very simple, as low as possible. As the optimal listening level for headphones is around 85dB, the blue line is the most relevant to real use. You can hear different levels of THD here to know what to expect from different amounts : https://www.audiocheck.net/testtones_thdFull.php

Next graph, also simple to read, expanded phase response/group delay :

​

This graph shows the delay of frequencies between eachother, as different frequencies can be produced at different times by headphones. This heavily impacts transients and imaging. The ideal curve is a flat line at 0s (or 0° for expanded phase response graphs). According to this research, the threshold of audibility of group delay depends on the frequency, but we can assume that anything below 1ms is inaudible (although this is debatable).

Next graph, the CSD/Waterfall :

​

This one shows the amplitude of frequencies over time. It will highlight any resonances produced by the headphones. The goal should be a shape as smooth as possible. It is normal for bass frequencies to resonate more as drivers move more and therefore have more inertia.

And finally, the most important one, and the most debated, the frequency response graph :

​

This graph will show the overall tonal signature of the headphones, as well as the channel matching between the left and right earcup. Researchers agree that this is the single most important thing when it comes to sound quality. The target curve for speakers is very simple : a flat line in a anechoic chamber. However for headphones, it's a bit more complicated than that because the sound comes directly at your ears, so a flat frequency response sounds unnatural to your brain.

The audio community and scientists don't all agree on a specific target curve, but the most popular one is the Harman Curve (here in blue). In short, Harman created this target curve by measuring the in-room response of reference speakers in a reference room with an average human head model, and adding a bass boost averaged by a panel of listeners to compensate for the lack of tactility. The response is smoothed (1/1 octave IIRC) to negate the fluctuations of HRTF between different listeners. Although this is the industry standard, keep in mind that some people prefer other target curves (including me).

4) So what headphones/IEMs should I get ?

Eventhough most headphones have a terrible frequency response, it can be corrected by EQ. As long, as your headphones have a low enough THD, you can EQ them to the target frequency response of your liking. The original frequency response should be smooth enough to avoid the EQ from causing too much phase distortion. You can use Sonarworks, Morphit or any tool that works for you. I have personnaly corrected my headphones manually with Crave EQ, as I personnaly dislike the Sonarworks target frequency, and its transparent mode uses linear phase filters only for the high frequencies (so latency is minimal, and there's no pre-ringing in the bass). Some manufacturers provide the measured frequency response of your own model, so you can be very precise with the correcting.

What if you can't use DSP ? You could look into IEMs. I personnally view them as generally superior to headphones when it comes to sound quality, because engineers are able to put multiple drivers to better control the frequency/phase response and distortion, and they're not subject to the reflections caused by your ears, which is the main problem with headphones.

Closed-back or open-back ? The real difference is not what most people think. The imaging is the same as the drivers are right next to your ears, so when you hear that open-back provide better imaging, it's because you better hear the ambience noise around you and it gives you the impression of a more natural imaging (it's all psychoacoustics). So the main difference really is dynamic range, as closed-back headphones lower the level of ambient noise. However many people feel better hearing their surroundings, so open-back headphones are far from being stupid. You can get the feeling of being in a confined space with closed-backs. Semi-open headphones are a good compromise that allow the ambient noise to reach you, but also allow the drivers to make a seal with your ears, often leading to deeper bass extension.


I'm not a scientist, so feel free to correct me on anything I said

 

Last edited: Feb 20, 2023

Very interesting, thank you.
I'm actually using AKG K712 PRO. Before them, I used Beyerdynamic DT 880 pro.
I feel that there is a great difference in the field perception, the space. AKG K712 PRO pro fell much more open.
My speakers are Event Opal, and the transition from AKG K712 PRO to Opal is good, I just notice a bit more of hi end with the headphones. But the bass seems equilibrated.
I work mainly on complex sound design, with advanced synthesizers, so there are a lot of frequencies involved from very low to harsh hi, and of course mixed and evolving.
I've also been mixing demos for my band, not looking for a super pro result, as the demos are to get live contracts, we play real funk, and old Soul and R&B. I could do it without a problem on the AKG K712 PRO.

I have (at least) 2 questions.
How do you find trustable analytic curves of headphones? I dont trust vendors.
Which headphones would you recommend?

Thank you

EDIT: I wrote the AKG K 712 Pro wrong at first, sorry..

 

Last edited: Feb 20, 2023

I was thinking of getting the Steven Slate VSX Headphones later this year.

 

Cool stuff, thanks for sharing. I am only producing and mixing on headphones.

What definitely helped me a lot is getting lots of feedback from different people with different monitor / room setups. Then I began to understand, where I should focus more or less (tonal balance wise).

My correction curve software is Sonarworks, and I am quite happy with it. But tbh, I am really using it less and less. It's only when I think, the mix has a pre-master condition, where I use Sonarworks.

But my go-to method is producing the entire track mostly (90 - 95%) in mono and when it sounds great there, I can almost be sure that it sounds even better in stereo. Luckily I have a superb mastering engineer who is really polishing my tracks every single time.

 

I did and it's been the best choice I made getting my mixing at the level I desire. It took e a couple of days and a lot of references to get used to the VSX, and then it took off in the direction I wanted. I have complete control over my mixes now.

 

thanks for sharing your experience.

do you have any recommendations for a good pair of headphones, as i was unable to find a headphone
i can trust for mixing.

 

One needs to be very careful, where to get the measurements from. I tried Crinacle with two different IEM's and adjusted their measured frequency response to their Harman target. One of them was brighter than the other. So I recommend to use the measurements from audiosciencereview. I think he is more constant.

You do not need a DSP hardware unit, just to add this maybe. It's more advisable to simply get the mathematically recommended EQ settings to adjust the frequency response to the Harman target and apply those filter settings to a eq-vst - doesn't really need to be FIR - that can create the correct filter curves and create a preset for this (Equilibrium creates exact curves in IIR mode, I don't know if pro-q does. One needs to check).

This phonitor from SPL is good. Who does not need to have all kinds of angles can simply use NovoNotes HPL2 Processor. It does basically the same thing. It's just set and forget. It is very advisable to use one of those binaural things.

The most important thing is: Yes, you can eliminate the room without spending thousands of dollars and have an absolutely clean listening environment. And it does cost not more than $400. (with headphone DAC and amp)
But it's still advisable to have a (untreated) listening room where you can check back the bass at the beginning. One will see by himself why this is important.

 

Last edited: Feb 20, 2023

Thanks for the input! I like what you said about speaker emulations, because they never felt good to me. If you use them, you should ALWAYS use them with your headphones, because otherwise you destroy your personal listening reference. It is crucial to know how music sounds on your gear, should be reliable as much as possible.

I definitely prefer mixing on monitors, but I like switching to headphones from time to time... be it as kind of a zoom or as a way to "beam into" the music... Listening on headphones can be really intense, which can be a thin line. Usually I love music being intense of course, but it can be fatiguing.

I have Beyerdynamic DT-1990. They have been the only headphones I really liked from the many I tried at the local music store. It's a bit strange, because they are quite dominant in the highs, which is usually not what I'm looking for. But for the way I use headphones it is a good match - they are revealing quite a lot of details.

 

I also find the DT770 to sound more open, I think that has to do with its frequency response that has less mud in the low-mids than the DT880.

Like @Lieglein said, Amir's measurements on AudioScienceReview are very trustworthy. DiyAudioHeaven also provides raw measurements and different useful graphs.

And regarding headphones recommendation, it depends on whether you can EQ them or not. If not, Dan Clark's Stealth and Expanse are regarded as the best of the best. They are 4000€ though, so more budget options could be the Austrian Audio HI-X60, Shure SRH440, or AKG K371 (my personnal favorite when recording, although they do suffer from channel mismatch).

If you can EQ them, there are a lot of good options. The only headphones I had that I couldn't properly correct were the Focal Elear and Fostex T50RP/T60RP. With the others, I could get an acceptable sound. However my personnal recommendations would be the OLLO S4X (S5X looks promising too), as I find them extremely comfortable and the brand can provide you with the individual frequency response measurements. They respond to EQ really well so you can get fantastic results. The cable on the S4X is shit though

I am personnaly using the Moondrop Variations as my reference system. They are tuned to a diffuse-field curve, compensated with a speaker in-room response curve and the Harman bass boost. I have no corrections on them as this is the sound signature I like best, but it's not the same for everyone.

I agree, Amir makes the most consistent measurements. I heard that Crinacle doesn't properly place IEMs or something like that.

Didn't know about NovoNotes HPL2, the demo sounds pretty good !

 

I'm very sorry, I wrote the name wrong... I meant AKG K712 PRO... Compared to Beyerdynamic DT 880 pro.

I corrrected my comment, very sorry ..

 

Last edited: Feb 21, 2023

thanks for the recommendations ...ordered the shure 440 headphone & probably later the akg 371.
i tested the beyerdynamic 990, hifiman he 400 and others before, and although they are praised everywhere
i felt none of them were suitable for mixing decissions...