Room Response Equalisation - a reflection

In recent weeks when trying to switch off thinking about work, I have turned my thoughts to Room Response Equalisation. There is much discussion about this in some parts from Linn and from REW users etc, but I was intrigued why sometimes the results can be really poor and artificial sounding when trying to compensate for room reflections…
I am also curious about speaker room coupling and why some speakers sound very bright in a room and others not…

So I started reading research papers. I found a particularly interesting scientific paper review of current thinking and trends on RRE: Room Response Equalization - A Review; Cecchi, Carini, Spors.

In short the paper goes into the RRE with a view of improving sound reproduction performance, and discusses the errors and pitfalls in some equalisation processes and speaker design. Also why sound performance changes with room temperature and humidity

I applied some of their observations and thinking to a response filter I created for my living room; essentially I took an average room measurement around the listening area; I then edited the correction (inverse response) to smooth out the adjustments, I shortened and simplified the filter by reducing the number of poles… and most importantly I did not compensate for the natural room / speaker coupling response which falls away in the high end and low end, around 100 Hz and gently from 10 kHz. (This is configurable in REW to experiment with)

The result was subtle on simple well recorded music… like a jazz quartet etc, or well recorded rock or pop, but play back an old recording or compressed rock recording and the effect was astounding … voices sounded certainly far more real and far more intelligible, strings sounded real, drums had tune … and no matter how compressed or old the music usually bounded along. We all enjoyed some old Ska recordings like we hadn’t before. Yes that initimate exaggerated Hi-Fi type affect is reduced on some recordings, but I conclude that is an artificial response anyway, and simply gets in the way of more complex recordings… where instead the realism and naturalness instead starts to dominate.

In a way I felt the response was similar to my vinyl replay… and looking at the response curve in places it resembled an RIAA curve for vinyl… so there might be something in that.

So the key was about effectively creating a smoothed inverse response, where the smoothing is proportional to the frequency, that, importantly, does not compensate for the natural room speaker coupling response which otherwise would be adding artefacts and making the sound unnatural or perhaps what some call ‘digital’

I do think some speakers are tuned to compensate for this fall off, which is why in some rooms they can sound too bright.

Anyway I thought I’d share… it’s certainly was a revelation to me, and a most beneficial and enjoyable one. I will post my filter response later. Might be food for thought with anyone wanting to try whose music replay is in a living space, not a recording studio.

4 Likes

That’s interesting, thanks for sharing. It’s something I’m minded to do but am technologically challenged in this whole area.

I’d be very interested to see your before and after graphs of the response curve.

.sjb

1 Like

Room temperature and humidity are to my mind important.

Listen to natural early morning bird song when the air is cold, relative humidity is high etc. It’s much clearer than during the day.

Not saying which is best etc, just that temp and R/H do affect sound.

2 Likes

Hi SJB, the graphs are hard to decipher - but essentially

You see the the room response in SPL with rise from 120 Hz and a gradual drop from around 1 kHz, the filled in turquoise colour is the edited inverse response. The light blue is the predicted response and the dark blue is the measurement. I have ensured the inverse prior to room response graduation is only subtractive and not additive - on the chart it is shown effectively inverted. You can also see in the lower mid through mid I have reduced room anomalies not removed them, and neutralised a sometimes a slightly forward sound in the room - more noticeable on compressed and poor recordings
Notice the room resonance at 48Hz which I have addressed to some extent, but is still there not that it is ever very noticeable - perhaps due to infinite baffle speakers - but is more problematic on ported speakers. Speakers are ATC SCM 19s

1 Like

Very interesting SiS. I’ve played with REW but didnt took the time to sit-down and do some proper analysis. What will happen to the big drops at 80 and 190 KHz? Are you able to smooth them out?

IMO, room correction is a bit of a no brainer - it’s relatively cheap and even free if you know what you are doing and have a recording device (and Roon or whatever to do the DSP). Even if you have to buy a microphone and get someone to do it professionally it is less expensive than lots of the audio tweakery we get up to, with the potential to deliver good results.

It’s on my to do list, but I’ll probably do the recording and then send the files to someone who knows what they are doing.

No I leave them, I only subtract not add, it really helps with the end quality. BTW I have expanded the scale. Good advice here from the latest thinking on room equalusation from the paper I refer to.

Linn’s Space Optimisation V2 model includes humidity and temperature parameters. SO V1 lets you print out PEQ settings, which I used to derive PEQ settings for JRiver MC (the settings are not the same even though I thought they should be, JRiver WiKi explains it).

My view is SO V2 is superb. SO V1 was merely good by comparison. I did try Dirac Live, but was disappointed - perhaps I didn’t pay enough attention to detail.

Did you do this in Roon?

I apply the fIR filter kernels for each sample rate frequency I use to Roon yes.

2 Likes

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.