What are my room correction options?

Jump to latest Follow Reply
von Chaps

von Chaps

Ars Centurion
1,909
Subscriptor
  • #1
    So, I have a Raspberry Pi 3B with a HifiBerry Digi+ HAT. I take optical out and feed a pair of Kanto YU6 active speakers.

    The Pi is running Raspbian and a custom build of the OpenHome renderer. The Pi runs headless. I have ssh access.

    This little system runs in a kitchen/diner which has, shall we say, interesting characteristics. There are resonances!

    Do I have any options wrt room correction and how might I go about it?
     
    von Chaps

    von Chaps

    Ars Centurion
    1,909
    Subscriptor
  • #3
    Oh, ASR. Not looked there for a while!

    Thanks. That thread is pretty close. It looks like Camilla DSP might be the thing. If I can get that working with my setup then, as you say, it's about defining the filters. I am not sure REW can be made to work on my headless Pi.

    Perhaps I need to set up an entire renderer on a Windows laptop including REW & Camilla and then try and export the resultant filters to the Pi. That sure is a rainy day project!
     
    von Chaps

    von Chaps

    Ars Centurion
    1,909
    Subscriptor
  • #4
    So, stop me if I am making a cock of myself....

    Because I like to understand what's going on, I generally try to do things the hard way first - before adding GUIs and automation.

    I installed REW and Equalizer APO on my desktop where I use MusicBee to listen to tunes out of a pair of Kanto YU2s. Because I don't have a mic yet, I used REW's sweeping sine wave generator and recorded the FR on an Android device using Spectroid. I then edited the Equalizer APO filter files to correct a notch at 82Hz.

    Yes, I know I'm way off piste here.

    But, the results sounded absolutely shocking and kind of what I expected from DSP. Honestly, I'd rather live with a resonating kitchen that this kind of garbage.

    Actually, having just typed all that, I can tell that I really am being an idiot. I'll leave it for posterity.

    Many, many people swear by room correction using DSP. They can't all be wrong. I guess I'll just have to do it properly! Mea culpa.
     
    von Chaps

    von Chaps

    Ars Centurion
    1,909
    Subscriptor
  • #6
    I've not got pics. I uninstalled the lot in disgust, with an intention to do it properly in the future. I might put it back together (it was pretty easy) and try again.

    As you say, the phone mic is miles from calibrated, but I did get a spectrum and I did see a difference when I edited the filters (by hand) in Equalizer APO. The difference was more or less what I expected to see but, as I say, it sounded like shit when I played music through it. Some sort of volume and possibly even tempo pulsing. Obviously hard to describe. But it did confirm my fears about inserting DSP into what I believe is a bit-perfect stream.

    To be clear, I'm not saying DSP is shit per se, I'm just wary of the side-effects and unpredictable interactions with everything else that's going on in the digital domain. At least with analog eq, you mainly just get phase variations and the ear is not sensitive to that.

    But DSP must work well, since so many swear by it.... right?
     
    yd

    yd

    Ars Legatus Legionis
    20,764
    Subscriptor++
  • #7
    I gather its great if done in a system accommodating to it.

    www.cardas.com

    Speaker Placement | Cardas Audio

    The Cardas Speaker Placement Guide. Setting up speakers in a golden ratio, golden cuboid, golden trapagon room, etc.. Speaker placement calculators.
    www.cardas.com www.cardas.com

    Personally, that is pretty much what I have used - which basically works well for bog standard rooms. Basically, as far out from the rear wall as your wife will let you get away with :eng101:
     
    • Like
    Reactions: continuum and von Chaps
    W

    w00key

    Ars Praefectus
    5,907
    Subscriptor
  • #8
    von Chaps said:
    I've not got pics. I uninstalled the lot in disgust, with an intention to do it properly in the future. I might put it back together (it was pretty easy) and try again.

    As you say, the phone mic is miles from calibrated, but I did get a spectrum and I did see a difference when I edited the filters (by hand) in Equalizer APO. The difference was more or less what I expected to see but, as I say, it sounded like shit when I played music through it. Some sort of volume and possibly even tempo pulsing. Obviously hard to describe. But it did confirm my fears about inserting DSP into what I believe is a bit-perfect stream.

    To be clear, I'm not saying DSP is shit per se, I'm just wary of the side-effects and unpredictable interactions with everything else that's going on in the digital domain. At least with analog eq, you mainly just get phase variations and the ear is not sensitive to that.

    But DSP must work well, since so many swear by it.... right?
    Click to expand...
    Well pretty basic room correction like Audyssey does more than just EQ. Basics are delays (distance) and phase match so speakers + subwoofer(s) are properly matched, and same for all the surrounds. It also does L and R different - my L is near a corner, R is on a flat wall, so the room effect on bass is very different between the two, left is more boomy, they get a different curve applied.

    And sometimes it is getting used to a new sound signature. Don't aim for flat, aim for a bit of bass boost a la Harman target curve, or V shaped for more "fun".

    Dirac does it better and in a very different way with impulse response correction. It's advanced voodoo.


    Trying to do that on a Pi though is hard though - I let my AVR do my room with its own calibration mic. But headphones are easy to fix with autoeq.app + EqualizerAPO. A must have for open headphones, those are very very light on bass.
     
    • Like
    Reactions: continuum
    W

    w00key

    Ars Praefectus
    5,907
    Subscriptor
  • #10
    von Chaps said:
    I run a pair of HD600s off another Pi. I had not thought to try and mess with that solution. I like bass as much as the next person, but I'm pretty happy with them.
    Click to expand...
    Try it on a PC / laptop. I have two pairs of Sennheiser PC360 headsets (HD 558 drivers + mic and volume control) and they gain much in clarity and fun on the Sennheiser Game ONE profile. Weirdly enough I prefer the Game ONE profile (line 2) over the HD 558 (line 3) one but as no one measured this specific headset, who knows what is more accurate.

    Accurate is overrated, don't need that for Daft Punk. ABC testing -> prefer option #3.

    1718119704591.png


    My other headphone, a HD650 is okay on its own but the correction profile adds 9 dB for the lowest bass, tells 200hz to chill and corrects the 10khz dip. It's much flatter than the cheaper headset.

    1718119821444.png

    Finally my much more fun earbud, SoundMagic E10, measures like this

    1718120083222.png

    No shit bass sounds good on it, it has extension down to 20hz (9 dB louder than the HD650!). But compared to the target curve, it has a huge hole around 2-5khz (flat=bad, HD650 follows target there very well) and too much mid bass. So again, EqualizerAPO to the rescue. As long as you don't hit excursion / power limits you can change the profile to what it should be, default only adds negative gain and never clips.
     
    • Like
    Reactions: continuum
    W

    w00key

    Ars Praefectus
    5,907
    Subscriptor
  • #12
    von Chaps said:
    You're talking about Audyssey now right? Does it come with profiles for those cans then?
    Click to expand...
    Oh nope headphones isn't something my AVR does. Autoeq.app has measurements for an insane amount of headphones and earbuds and let you export that to settings for a ton of phone and PC apps, and also some devices like MiniDSP.

    I use it with EqualizerAPO on Windows.


    AutoEQ can also export to standard 10/31 band EQ. Pretty much any system has that right?
     
    • Like
    Reactions: von Chaps
    von Chaps

    von Chaps

    Ars Centurion
    1,909
    Subscriptor
  • #13
    So, I thought I'd strip this back to basics so as to understand the process. I failed.

    I installed EqualizerAPO on my Windows PC. I then loaded the HD600 profile from Autoeq.app. I played it back through the Kanto YU2s not the headphones. I expected the EQ to be not necessarily ideal, but it should work, right?

    It sounds horrendous. There's the same chugging, pulsing, tessellation of the sound, that I mentioned earlier. Even if I scrap the Autoeq.app-generated config and just create a few filters by hand in the EqualizerAPO editor, it sounds crap too.

    I am feeding the Kantos via USB and letting it do the D2A. My audio is FLAC files ripped from CD - 44.1. Is there anything obvious I'm doing wrong?

    Honestly, if I can't get this working with all the tools available on Windows, I feel I don't have a hope on Raspbian.
     
    von Chaps

    von Chaps

    Ars Centurion
    1,909
    Subscriptor
  • #15
    I don't actually have analog out (it's a laptop, so I'm not considering the speakers), so it's going to be running though some external DAC at some point. I had wondered if the DSP is interacting with the DAC in some way, but that would not just be a "me" problem.

    AFAICT, running only a preamp (with moderate aadjustment) seems OK. I think I need to learn a bit more about what is and isn't possible with a setup like this. I think I discovered that attenuating frequencies is better than trying to amplify them.

    Perhaps I just need to spring for the mic and come at this properly. I am not having much luck with my hobbled together approach.
     
    W

    w00key

    Ars Praefectus
    5,907
    Subscriptor
  • #16
    von Chaps said:
    AFAICT, running only a preamp (with moderate aadjustment) seems OK. I think I need to learn a bit more about what is and isn't possible with a setup like this. I think I discovered that attenuating frequencies is better than trying to amplify them.
    Click to expand...

    Never ever boost signal, that's asking for clipping. If you need to boost bass, you reduce everything else instead. EqAPO shows > 0 dB in red, that's bad. If you do have want to boost, use it with a negative preamp.

    Example: highest is 0dB and only goes down.

    1718119704591-png.82882
     
    von Chaps

    von Chaps

    Ars Centurion
    1,909
    Subscriptor
  • #17
    Yea. I think I had concluded that. I hate to throw away s/n, but that's probably irrelevant in reality.

    In principle, +db should be acceptable in some cases, but it's impossible to really know. And how that interacts with Replay Gain 🤷‍♂️

    I suspect this may be where I am going wrong, but it's hard to know without any indication of whether I'm approaching clipping (in the digital domain).
     
    You must log in or register to reply here.
    Jump to latest Follow Reply
    Jump to latest Follow Reply