Unfortunately SO wasn’t as problem-free as I hoped for. I used SO exclusively for almost three years with my Linn ADSM/3 + ATC actives setup. While it helped with room modes, I also lately got this feeling that it takes too much out of the music. Few weeks ago I borrowed UMIK-1 and installed REW to do some measurements in my space. I also measured how SO affects the sound. While it corrects some problems, it didn’t fix the most obvious ones and also substracted too much from the bass making the sound too thin and lifeless.
With UMIK-1 and REW I was able to pinpoint the one bad room mode at ~36hz and fix that and also correct another slight peak between 65-75hz. SO wasn’t able to fix the room mode at 36hz properly even with heavy tweaking from the original settings. With REW, I made convolution filter based on the measurements and eq of it and uploaded it to Roon. My system has never sounded this good. If you can borrow a mic like I did, this is a completely free upgrade. Highly recommended!
Linn SO may have accurate measurement data for some speakers, but not all of course, and then simulates the room’s acoustic behavior to calculate corrections.
But as good as that modeling may be, it’s never going to capture the picture as accurately as well made in situ measurements.
Thank you. Just for clarification, how do you determine “the averaged measurement”?
I’m just realizing that the graphs I posted do not actually accomplish what you were suggesting here. For that I would need a graph which shows 1) The predicted curve from the initial measurement’s EQ window, and the 2) Measurement curve once the filters have been applied. If that’s correct, then I would assume that could only be done in the overlays window, where I see"All SPL" in one tab, and “Predicted SPL” in another, but no way to combine a measurement curve with a predicted curve, in one graph. IOW, how do you go about comparing predicted to actual?
You measure and apply, as I suggested, psychoacoustic averaging, then set target to be below the smoothed curve and generate filters while maybe having to dismiss a REW level warning - badabing badaboom
Just had Thierry @Home_Audio_Fidelity create me some new filters for my office. Its not a gret room acousitcally at all, even to talk in plenty of slap echo and the valuted pitch roof doesnt help and limited soft furnishings. But wow what a difference, its never going to be the best system as its a lowly pi with Hifiberry Amp2 HAT which sounded better than i thought it should but now its even better so RC can help even the lowliest of systems. Also got updated main room ones which not had chance to try yet.
Nicely done, but what are the assumptions here insofar as plugging the PC output into a stereo system? Does this assume that Roon Core is running on that PC and the output is fed to a DAC first? What if Roon Core is running on Nucleus? Can this filter be deployed on it?
Any filters (including HAF or REW) generated for Roon’s DSP get imported into Roon by a Windows or macOS remote. It doesn’t matter what the Roon Core is running on.
From REW’s signal generator dialogue, export the measurement signal as a .wav file and import it into Roon to play it via your regular signal path, while measuring via mic on PC with REW.
Either use pink noise for just frequency domain, or a sweep with timing blip to also extract time domain results.
Of course, you need to use either a Mac or a Win PC to run REW.
EDIT:
And as @glc650 says, generated filter .wav’s zipped up for all sample rates then are imported from Roon’s convolution dialogue.
I have a Pink PN noise imported in Roon, which I use when doing measurements. That way you get the exact same audio chain for your measurements as when playing music. If you want to use sweeps instead, it’s also possible to play from Roon as long as you have the “chirp” sound marker.
One thing I have wondered about though is if HQPlayer filters and up-sampling should be in use when doing measurements, or if it’s better to disable them. I imagine it doesn’t matter much but maybe @jussi_laako knows?
EDIT:
I’m not Jussi, but…
If you think they make an audible difference for you, you should include them, although I’m pretty sure, their impact will be absolutely dwarfed to, at least, invisibility by speaker/room influence.
Another shout-out for Thierry @Home_Audio_Fidelity, who created some filters for me - I was already quite happy with my self-made filter based on the instructions in this thread, but his filters really improved the sound further! Better instrument separation and localization, and especially vocals got a wonderful, silky quality.
@Marin_Weigel
This is interesting. Can you please help by giving more details of generating the Pink Noise sweep along with a timing blip. I have not been successful in this. Many thanks
I just tested this guide and it gave me excellent results, so it’s a good alternative to this guide with RTA and moving-mick. It’s also pretty easy and fast to do (no complicated FIR EQ filters involved, and only 2 sweep measurements, make the sweeps long though).
All this talk about room correction, and not one word about the first and correct way of doing it… With physical acoustic room treatment.
Besides, running digital room correction through Roon is useless when you also listen to CDs, tapes and vinyl. So you make your streaming sound marginally better and say the heck with everything else in the system. Hmm…
To get the optimal sound you either need to be in a specially built studio (nonparallel ceiling-floors and walls) or as most of us do combine as much acoustic treatment as possible with digital room correction to fix the remaining issues.
Not sure about that “vinyl” or “tapes” you mention, maybe I am too young to know
Again, you’re ONLY correcting your system/room for one source. For people who have and use multiple sources all the time, this version of room correction is pointless.
And as I said, physical acoustic room treatments is and should be the first line of defense. There’s things that no amount of DSP can take care of.
