Show HN: HiFiScan, a Python app to optimize your loudspeakers
github.comI'm not an audiophile in the obsessive-compulsive sense, but I've been recording music in my home studio for 20 years and I know my way around it. This sort of calibration is not ideal. Not only are you measuring with a device that has an imperfect response curve, but you are also measuring the room at a single monophonic point in space. The way that sound interacts with the room and your ears is far more complex than that. Ultimately, this is a bandaid for a poorly treated room. If you're serious about getting a flat response curve from your monitoring room, you're far better off learning how to treat the room properly and how to position your monitors within the room for the best results.
Calibrated (my Audyssey is +-1DB 20hz-20khz) mics do exist and calibration files do exist for some mics, plenty good enough considering how terrible the average house/apartment room is.
Your totally right about room treatment and why I don't spend more than the budget end of HiFi, because I know how terrible my room makes everything sound..
Ive measured a few rooms and sets of HiFi and the biggest issue I find is not the general frequency response, but room modes, they peak far more than even the frequency response from a cheap boom box. Without active EQ or a notch filter and extensive room treatment then trying to flatten out the response of a speaker is futile if you ignore the room modes. Add in transient and phase responses and you add additional challenges to getting good audio.
Have you tried Dirac Live? Particularly with a receiver that supports their Bass Control feature? My NAD receiver only supports regular Dirac Live and it's still much, much better than any integrated room correction from Denon/Marantz, Yamaha & Co.
The improvement is so noticeable that I assume it would take a lot of resources to fix the room physically in order to achieve similar results.
I don't want to sound to enthusiastic but I'm pretty certain I'll never buy a receiver that doesn't support Dirac Live.
It does however look great for helping me characterize differences in my 3D printed headphone designs. My old resin-printed design has incredibly good bass, and when I changed things up for FDM printing I lost all that low end. I need to try adjusting my design and quantifying the results, as I love the sound of my resin-printed design but would prefer to move to FDM designs as they are much easier to print and make.
Here's an image comparing my old headphone design with good bass in blue, and the new easier to make design with bad bass in red:
https://twitter.com/TLAlexander/status/1569181219446980608
The 50mm headphone drivers were globally out of stock for a year so it is safe to assume the new drivers are from a different manufacturing batch. But my hope is that the problem lies in my headphone design, and the new and old drivers are roughly the same. I don't know anything about headphone design and seemingly got lucky with the old design, so if I need to adjust my new design that's not a big deal. But if the new drivers aren't performing like the old ones, I may not be able to fix things. So next step will be to swap drivers from the old and new headphones and see if the problem follows the drivers or the headphone design.
Images of the old design as well as the link to the onshape files on this git repo:
Measurements (particularly, FR and distortion) aren't there or are hard to find.
It'd be most interesting to see them.
Yes I always wanted to do them. For the original set I just compared them to my friend's expensive headphones. It was only today that I did any kind of frequency response testing. But since I don't have a calibrated microphone, it is more useful for doing relative comparison between my headphone designs than it is for providing absolute numbers.
What I wanted to do was build a little headphone test dummy head. But for my relative tests I just shoved my podcast mic, Audio Technica AT2005USB up to the center of the driver on one side of the headphone. Do look at my linked tweet for those FR curves.
However this little adventure has encouraged me to consider building the little head dummy. I have some nice little capsule mics I was playing with that would work nicely, and I have a datasheet for those that probably includes a curve.
This hobby work of mine goes slowly, but I appreciate your suggestion. I will try to tune up my FDM 50mm headphone design so that it sounds at least as good as my original resin printed design, and then look at building a better characterization rig.
> you're far better off learning how to treat the room properly
Would you have some pointers on this?
I've been looking into this and while I've found pointers on "what to do", what's missing is where to actually find the necessary panels and how to figure if they're actually worth anything.
Here's some basic before/after examples that might be useful. This is kind of a deep rabbit hole. My dumb brain still dreams of blackbird studio c every once and a while.
Build your own panels with rockwool insulation. There way better than almost anything you'll find on the market and easy even for no talent carpenters like myself
Exactly. If you really want to get into it the depth of the construction of the panels you need is based on some maths - density of the insulation material and it's particular properties but all of that can be found out on forums such as this one: https://gearspace.com/board/studio-building-acoustics/ - vs the frequencies you wish to treat.
You can put your room dimensions into a calculator and get a rough idea of some of the try and find the modes of the room - which you want to treat, though sometimes it takes trial and error as well. But you want to treat the point of first reflection and then have bass trapping in the corners.
Don't buy that "acoustic foam" that looks like egg cartons, it's rubbish.
How do you cover those? I'd expect drywall or similar would negate most benefits.
Fabric that you can breath through easily for looks. Under than you can be very thin fabric that makes sure that the fibres from the insulation doesn't escape.
I use this for looks: https://www.camirafabrics.com/en/contract/inspiration/acoust...
I've found https://ehomerecordingstudio.com/acoustic-treatment/ to be a well-written and informative guide.
You might look here: http://realtraps.com
You can probably reduce some room resonances.
Counter-point: The use case is making relatively-shitty speakers in front of the computer better, and single point in space isn't really that much worse than trying to somehow put speaker in place your ears are and accomodate for all that.
> Ultimately, this is a bandaid for a poorly treated room. If you're serious about getting a flat response curve from your monitoring room, you're far better off learning how to treat the room properly and how to position your monitors within the room for the best results.
Well, doh, but it costs zero dollars and very little effort.
You'd also be relying very heavily on the microphone used to measure it.
That's the 'device with an imperfect response curve', I assume.
In fairness, the readme does state:
> A good microphone is needed, with a wide frequency range and preferably with a flat frequency response.
By 'preferably' I assume it's implied that it can curve-fit (whatever's needed, I know next to nothing about this) to a non-flat microphone response, as long as it's known, but if it's flat then no need.
If it's unknown (and non-flat or assumed non-flat because it's cheap and doesn't make any claims about it) then that's the real problem, no point trying to do anything because it's like trying to construct a level floor with a shoelace for a spirit level.
i always wondered, is it possible to take a cheaper mic or iems and "flatten" them via an eq, to perform nearly as well as professional gear that's 3x the price?
i just picked up a pair of KZ AS06 iems [1] and my listening preference is U shaped (which is how these are dialed in out of the box), but i imagine with quality hardware and e.g. 3+ dedicated, drivers it should be possible to flatten them out in an eq.
[1] https://old.reddit.com/r/headphones/comments/eqpsen/kz_as06_...
Yes, to an extent, and this is frequently done to great effect. However frequency response isn't everything, there's also e.g. group delay, off axis response, and harmonic distortion. In particular, boosting response in areas a speaker is deficient often causes a huge increase in distortion, so you have to balance.
Yes, but no, but yes.
AutoEq (https://github.com/jaakkopasanen/AutoEq/) does something like this so... yes?
But targets various HRTF curves which aren't neutral or flat, and they usually aim for a Harman target of sorts, so no...
But there are some neutral-sounding curves that seek to just emulate pinna gain, so they have a ~3KHz peak which is required to 'emulate' the natural resonance of your ear, which you don't get naturally when you jam a driver all the way into it.
There's no AutoEq preset for the AS06s. Here's a representative sample targeting the 2019 Harman in-ear target for the KZ AS10s for funsies though: https://github.com/jaakkopasanen/AutoEq/tree/master/results/...
the AS10 and AS06 curves look quite similar, so might work, thanks!
https://www.reddit.com/r/headphones/comments/afsro3/review_o...
Commercial systems that do this kind of room correction generally have a limited range of recommended microphones, and the more expensive (and hopefully better) ones will have the microphone calibrated and factored into the room correction - for example, Anthem's Room Correction (ARC) ships mics that have a serial number. You plug that into the ARC software, it looks up the factory profile of that specific mic, as it was recorded at build time, and weights the calibration for it.
Here's a measurement standard mic:
And what would be the "80% of performance for 20% of the price" equivalent of it ?
Minidsp umik-1.
This type of software, is just a bandaid and really doesn't work very well (though it can work better with headphones). Properly thought out and tuned acoustic work is what is needed.
I am lucky enough to have a spare room in my house, and set out to build a studio (an almost life-long dream) and decided that I didn't want to compromise on the acoustics and spent some time looking into the subject. In the end I built it myself with a huge amount of acoustic treatment (lost a large amount of the volume room), but more that that I enlisted the help of a professional who could do the maths and help with not just the trapping but also the panels that are needed. In the end after I built it was also tuned with DSP by the professional, has what you would normally call 4-way speakers with the subwoofers going to a higher frequency than most would consider normal and even the desk was specifically chosen to not cause a problem for the listening environment. The difference between this and something like Sonarworks (commercial software that I tried for a laugh beforehand) cannot be overstated. It's basically flat between 23hz (slightly rises at 20hz I believe) and 20Khz - we actually tuned in a more natural response curve.
It's still a home studio because it's in my home and I don't do anything commercial with it, but it's pretty much mastering grade, all with materials that are available in a builders yard and the special sauce, someone that knew what they are doing. Not everyone has the room or space to do this, but most people can build some bass traps and something to tame first point reflections.
This project still has a good bang for the time or money buck.
Life has compromises. You do give up some things to build a perfect studio.
There is an older project with better math inside: http://drc-fir.sourceforge.net/
For starters, it doesn't try to achieve a phase-neutral response, because a phase-neutral response created in a room is only valid in one point of the room, and creates pre-echo artifacts elsewhere. In fact, it tries to separate the response of the speaker itself from the response of the room, by setting a threshold in the time domain, so that everything coming before it must be unaffected by the room. Then, everything coming before the threshold is corrected to a linear phase, while everything else is corrected to the minimum phase (thus making the second part of the filter purely causal).
Also, they provide an argument, citing literature, that equalizing to a flat frequency response would be wrong in a room, and thus provide an option to remove excessive treble and achieve a 1dB/octave roll-off.
Please see the details at http://drc-fir.sourceforge.net/doc/drc.html
> because a phase-neutral response created in a room is only valid in one point of the room
Author here. The term "phase-neutral" simply means here that the impulse response is symmetrical and doesn't add a phase shift. It doesn't even try to neutralize the phase characteristics of the room, which is what you may be thinking. In fact the phase information from the measurement is completely discarded. Furthermore, the frequency response is averaged to get a more general and robust (less over-fitted) correction that works pretty well across the room. Try it...
Well, if you discard the phase of the original response anyway, then you can shave a few milliseconds of latency by switching to minimum-phase (which is causal, not symmetric) instead of linear-phase. The math is in scipy.signal.minimum_phase.
Newbie question: how do we know we can trust the microphone?
It sounds like a chicken-and-egg problem to equalize speakers with an equalized microphone, but maybe microphones are simpler and can be assumed to be equalized ?
You’ll need a calibration curve for the microphone. Even of the same model, there is a lot of variance.
MiniDSP makes some calibration mics that run about 60 bucks. I used them as a cheap instrument for some lab work where I needed a calibrated mic a while back and was very impressed with their performance for the price. They ship with a little code that you can use to retrieve the calibration curve from the factory, and I know a lot of people use them for hifi calibration with REW.
Anyone interested in this area should also know that above ~2 kHz it doesn't matter what you do for magnitude equalization because you'll be dominated by sub mm variations in position and direction. The only way to get any amount of repeatability above 2 kHz is with IEMs.
This is dealt with by smoothing the spectrum in a way to preserves the power density. The constructive and destructive interference then cancel each other out.
What does "smoothing the spectrum" mean? What operations are being performed?
I'd guess: Fourier transform, a power density preserving blur convolution, then inverse Fourier transform.
But I am not familiar with the field of signal processing.
That's my guess too but my hunch is that would sound like absolute crap. Intentional IMD. Something to experiment with.
There’s an example of doing this on the readme- scroll way down
There are cheap calibrated mics available. There's one for about $20 from Dayton Audio.
A microphone that is linear to a dB or so is far cheaper than a speaker that is linear to 6dB and room treatment that maintains that.
Not an audiophile, but one way might be tuning forks. That said, I would be super surprised if this was needed for high-end microphones.
A microphone’s ability to reliably identify a frequency is excellent, even if the microphone is cheap, crappy and uncalibrated. It’s almost entirely a function of whatever clock is used to digitize it, and oscillator chips that are just fine are ubiquitous.
The issue is calibrating the amplitude response at a given frequency, and a tuning fork won’t help.
edit: those quartz oscillator chips have a lot in common with tuning forks.
Now I'm imagining using thousands of tuning forks with calibrated knockers.
I wonder what are the differences between this tool and industry standard REW app - https://www.roomeqwizard.com/
Source code, for one.
This looks cool. I'm not sure if they are intending to go all the way to room correction but it can really do wonders. A good while back my music setup used filters calculated by an open source FIR tool with playback driven by an older version of Shairport (emulating an AirPort express) using BruteFIR as a convolver. Fiddly to set up but it sounded really good.
1. http://drc-fir.sourceforge.net
I'm a psychoacoustician and this is not the way, very sorry to report. Others have touched on the acoustic issues already, so let me touch on the psychological ones: your perception of sound from loudspeakers doesn't just depend on the acoustic waves hitting your ears. It also depends on your personality and expectations. If you genuinely believe that doing a seance to drive out the poltergeist from your speaker set up will make the sound better, it will be difficult to convince you otherwise precisely because the acoustics did not actually perceptibly change.
A friend loaned me a fancy usb DAC a while back and I used it to listen to music while I worked. After about a day or so I asked her if it was my imagination or if the audio really did sound better. Her answer was that there's no difference between those situations: if I imagine it sounds better, it does sound better.
I don't have to use my imagination to hear the signals coupled into my USB DAC when I move my mouse. They're really there and I really don't want them to be.
Depends on the USB DAC, and maybe/probably the PC. I don't have/hear interference with my current setup. I did have the issue with another DAC; IIRC I put an RC filter into the USB cable power lines (or did I just add an R into the ground line? Do some research if you plan on trying this).
PCs these days often still have an optical toslink, that can be used to avoid the issue.
I have an ODAC+O2. They're not easy to source for cheap these days. My issue is that the power supply filtering inductor has a cracked iron core (due to me being clumsy). The cheapest solution would be to replace it, but it's not super easy to swap out surface mount components.
I'll just deal with it by keeping the O2 volume low and cranking the volume on a second amp. Long term I'll just buy an element or a schiit.
If you use a second amp, why bother with the O2 at all? Or do you have the integrated ODAC?
I've built the O2 as well, and loved it, since for the first time I could drive the HD650; it even has enough power to properly use them on a flight (don't do it, you'll go deaf). But the form factor just isn't that great for on-the-go. Plus with the batteries installed there is no space for the ODAC, requiring a DAC solution (or a phone with a good DAC).
Give the FiiO btr5 a try instead of spending too much on a schiit, or look for similar devices from other manufacturers. A lot of them can do both USB (96kHz/24bit) and decent (LDAC) bluetooth. I could not tell a difference between a friend's BTR5 (dual ES9219C directly driving the headphones) and my Q5 (dual mono akm4490, ad8620 gain stage with dual op926 buffers when used balanced). Of course my hearing could be too damaged^^'
The ODAC is soldered to the O2. The O2 is in pristine condition (and I did splurge for an OP2227 in the voltage gain stage and WIMA caps). The analog in works on the O2 so I may abandon the ODAC in place.
The Fiio options do look good. My current desktop application has me switching headphone and mic between a work laptop and personal desktop and headphone and speakers often. I use analog switches, but I'm leaning towards a one box solution in the future rather than adding more hardware. So something like a mayflower ARC mk2 perfectly covers all of my demands (including a bass boost option for my open-back headphones). I already use a USB switch with a powered USB hub, so really it would be swapping out 3 boxes and many cables for one box. The schiit hel 2 looks really great but I've read reports of 200 ms latency.
It's an unwritten rule in the studio scene around me to have a specific fader that is prominently placed but does nothing. To use when certain musicians (usually guitarists) demand to raise the gain of their instrument into unreasonable territory. Seems to work reliably to look at them and very slowly raise that fader until the they say it's good
"Frequency response does not matter because other things also matter and I have a PhD in these other things." This is a complete non-argument.
If you read again carefully, that is not what I said.
> If you genuinely believe that doing a seance to drive out the poltergeist from your speaker set up will make the sound better, it will be difficult to convince you otherwise
I think you just found the next big thing in audiophile fads
I'm more interested in inviting the right kind of poltergeist to dwell in my speaker set up, to give playback the warm paranormal sound that is clearly missing from sterile exorcized speakers.
Making sure to keep things as equal length as possible, run your speaker cables so that they form a pentagram inscribed in a circle. 5 speakers only, although a subwoofer is fine.
I genuinely wasn’t sure whether psychoacoustician was a cheeky synonym for audiophile or not. :). I looked it up and, sure enough, psychoacoustics is a legitimate field of study.
Nonsense. Not everyone is an "audiophile" (in a bad sense) who believes in silver speaker cables.
I have made a small webtool to help calibrate various EQs by ear. It kind-of mimics a graphic EQ in the browser which can also play tones around the EQs frequency bands, which should sound about the same loudness as their neighbors according to the ISO loudness curve. I increase or decrease my laptops EQ bands until the tones on the webtool play without obvious difference. This is sure to be an unsatisfactory process for technical purposes, and I couldn't even guarantee that I implemented the loudness curve well, but I have a lot more success using it to help tune EQ than without it.
Isn't it basically what "DRC" does? (Digital Room Correction)
https://en.wikipedia.org/wiki/Digital_room_correction
I don't remember the exact order but way, way, way before the $10 K USD digital audio cable snake oil, audiophiles are going to say that DRC is the second single biggest thing that can enhance the quality of your setup (the first one being which speakers you're using and how you place them). Then source quality/amp/dac. And only way further down the line, for those who believe in voodoo, $10 K digital audio cables.
An objectivist audiophile would say that room correction is among the three or four grossly consequential parts of the audio chain. They are, in serial order:
0. Source material
1. Room correction DSP
2. Speakers (including subwoofers and crossover configuration)
3. Room acoustics (including positioning of speakers and listeners)
4. The human (ears, experience, expectations, ego, etc.)
All of the above are more consequential than anything else, assuming the core components are not total garbage, underspecified or malfunctioning. This includes the DAC and amplification.
Of the above list, I would place room correction at the bottom. (That still places it well above many things subjectivist audiophiles obsess over!) It is the cherry on top of a great system, not the means to achieving greatness. And it lets you get away with some things (most notably, mismatched speakers) to a greater extent than otherwise. But despite the name it can’t fix most real acoustic problems.
It can also make a system sound worse if it’s not used properly.
Seems to be some kind of DRC.
$10K digital audio cables are never a good idea.
I remember I went to some audiophiles house once to demo some speakers, and his "hobby" seemed to have taken over the house and common sense. He had crazy expensive audio equipment and some of the thickest cables I have seen, with the cables all suspended on little bridges.
All this in a room which was basically a square brick construction with glass windows on 3 sides, no thought to any treatment. He didn't seem to understand that the room was effecting the sound more than any DAC, Amp, Cable, or any of the other voodoo that was going on. I couldn't properly demo the speakers because of a particular standing wave. I concluded he probably had a hearing problem, he concluded he needed to upgrade a cable.
I’ve tried this for my speaker setup. And the problem is that the frequency response is a function of volume. For example, the louder I play music the more the bass is accentuated. I think this is because of standing waves.
So the problem I find is that when the volume is low the bass is too low, and when the volume is high the bass is too loud. Only when I play at the same volume as the equalization was performed at do I get a good result.
This is a common psycoacoustic effect and probably not due to the loudspeakers or the room acoustics (which are linear): the (perceived) loudness of a tone is frequency and level dependent [1]. This makes sounds more bassy at large sound pressure levels.
Loudness is also distance dependent, as in, not just how far away the sound source literally is, but how far away you think it is.
Some hifi systems have a "loudness" setting that raises bass+treble to compensate for this effect at low volume.
Asking because I'm not smart enough: Is this kinda similar to what the Sonos Trueplay feature does? (Where you move your phone, and/or a mic-enabled speaker itself, around the room so that it plays and measures various frequencies to calibrate)
That one is an automatic room equalization, it's different from speaker equalization and it probably should be done after speaker equalization. But it's more useful for the end user.
Sorry, I might've linked the wrong patent then. (I meant to, but failed to, find the one that handles speaker equalization for a single speaker).
By room equalization, do you mean normalizing volumes between different rooms, or...?
Room equalisation is about counteracting the effect of the room and the placement of the speakers. For example, when a speaker is close to the wall, or in a corner, low frequencies are amplified, even if the speaker on its own has a flat response cureve, so you would reduce low frequencies to adjust for the room.
I was thinking the same thing myself, if this was similar.
This is very nice. I also appreciate the pointers to various equalizer apps in the README, I didn’t know a couple of them.
I'm using the commercial https://www.sonarworks.com/soundid-reference and it's amazing.
I'd say the worse your setup (especially your room) the more magic it does.
I did it without an individually calibrated mic though (but with a decent measuring one), wonder how much better it could be.
I'm also using this.
The results are very good. I have studio monitors and a crappy room setup, and the calibrated sound is much better. I purchased the kit with the supplied mic.
That said, the software is unstable. To the point of uselessness. It caused so many system crashes that I - very sadly, because the results are so good - just don't use it anymore.
Hoping they fix stability in later versions so I can go back to using it.
I think a lot happened shortly before the rebranding to "SoundID Reference". If you tested before that, maybe give it another go.
I stopped using it about a week ago.
Is it affordable for mortals or is this a business only offering?
Certainly expensive. ~300 eur/usd with mic (and you need a proper audio interface to support the mic).
A bit sad, because it might do most for less expensive speakers and untreated rooms.
It's cheap compared to a Trinnov system. But honestly you might be better off spending 300eur/usd on wood, Rockwool and some fabric.
So many negative comments here. I for one think this is absolutely fantastic tool to calibrate cheap television and computer speakers so they don't sound like complete crap. We're talking about almost 20dB variations on the spectrum here.
Cool project! I recently bought a set of iLoud MTM monitor speakers which come with a special mic which they use to analyse the room and correct for it in a similar way to this.
It makes a good difference to the sound - highly recommend the speakers if you are looking for a smallish set of monitor speakers that sound great and can be used very near field so you can use lower volumes.
Yeah, those are kinda cool. They have the added advantage that they don't have a huge amount of bass, though what they do have is impressive for their size to be fair.
But because of their size they don't always activate room acoustics in a crazy way, and a lot of people monitor with them fairly close so don't need them loud either further lessening the problems.
> They have the added advantage that they don't have a huge amount of bass, though what they do have is impressive for their size to be fair.
Yeah, this is why I got them really. I have a pair of Event 2020 BAS v3 but our new apartment is much quieter than the old one and I'd feel self conscious blasting them out, so was looking for something which still sounded good but worked well at low volume. I'm pretty impressed with the bass to be fair. I also tried the smaller model (iLoud Micro) but for the sort of music I like for which bass is reasonably important (techno etc.), they weren't quite there. Still impressive though
> a lot of people monitor with them fairly close so don't need them loud either further lessening the problems.
This simple insight is gold. But is it actually true? The standing wave should still pop up. Though with less energy it's probably mostly handled by the furniture.
Whatever the case, having monitor speakers sitting close avoids/lessens the issue of the first reflection point, making the higher bands sound much less "muddy". This can be improved by picking a speaker with a strong beaming characteristic. Eg 4" broadbands will bundle the acoustic wave quite strongly in the higher frequencies. Sounds muffled for bystanders outside the beam, but amazing stage and resolution for the one or two persons inside of it.
How does this compare to dirac?
I'm finding the "this is a horrible idea" responses amusing. I don't know if there's something fundamentally different about the way this project works versus Dirac/XT32 or if the naysayers aren't familiar with it. Or maybe there's an anti-room correction sect of audiophiles that have remained hidden to me.
Not 'fundamentally', but using one single point would probably be the main issue. Move the microphone 5cm and the response measured is going to be different. Dirac and manual methods with REW use multiple points and/or just moving the microphone around.
I think it would be cool to make a more advanced version that corrects for many types of nonlinearities: amplifier distortion and mechanical parts resonating badly.
> A good microphone is needed
What qualifies as good enough?
Is it worth trying with consumer mics like the ones built ins on phones and laptops?
I highly suggest getting flat neutral speakers first. Preferably high end studio monitors. What would be interesting is if someone can work on music-specific optimization based on a handful of inferences and ML.
Nope, it's better to treat the room first, then invest in some quality monitors like Neumanns. Basic budget monitors will do great in treated room.
How does this compare to just buying a good headphone?
Headphones should also be calibrated by doing EQ. See https://www.reddit.com/r/oratory1990/wiki/index/list_of_pres...
A good headphone can probably still outperform a speaker system. The tradeoff is that you have to wear a headphone. In my case, I just hate them. It's just more pleasant for me to listen from speakers, despite the fidelity tradeoff.
Try good half-open cans like Beyedynamic DT880. I can wear them all day.
I never really liked wearing headphones as well. I've setup my room such that I can use the big stereo speakers plus a decent mic (Samson Go Mic) for voice chats.
However, Sennheiser HD650 are a pleasure to wear. Even for longer periods of time. I use them with a bluetooth+USB DAC/amp (Fiio Q5; outdated) and a short cable; so I'm pretty flexible how I can use them.
There's also the fact that when you really get "into" the music and start moving your head, there are basically two scenarios: you either get the "hum" of the headphones moving on your ears, or the headphones stay put but the clamp is so tight that you can't stand them for more than 10 minutes.
Headphones are excellent, but I find I have a different experience with headphones compared to music coming out of speakers. Sometimes I prefer one over the other.