Automatic Room Correction has won the Subwoofer Wars

Imo room correction works great in the subwoofer region, but not so much further up the spectrum. This is because the microphone picks up the reflections along with the direct sound, and does not differentiate between them. That’s fine at very low frequencies where the ear’s time-domain resolution is poor, so the ear is likewise not differentiating between the direct and reflected sound.

But further up the spectrum the ear differentiates between the direct sound and the reflections, perceiving and processing them differently. And in general it would be a mistake to "correct" the direct sound based on the contribution of the reflections. I’m not saying that improvement via EQ isn’t possible further up the spectrum, but imo it should be done by hand, not "automatically" based on what an ungated microphone captures in-room. 

Latency might be an issue with some digital room correction systems, I suppose it depends on the specifics.  

Duke

Dannad wrote: " ARC can detect accurately speaker distance based on first time of arrival and based on timing could differentiate between direct and reflected."

It depends on the wavelengths and reflection path lengths. In order to get good direct-sound data at a given frequency the time window has to be open and reflection-free for at least one wavelength, and possibly more. If the wavelengths are long relative to the reflection path lengths, the reflections start to arrive before the first wavelength is even finished. Now factor in the time-delay-induced phase rotation of that arriving reflection and we have bad data, because the microphone can ONLY sense the combined net pressure at its location - it CANNOT tell how long a path some or all of that pressure (or lack thereof) travelled before arriving.

I use time-gated measurements regularly, and in a "normal" room there will be a lower limit, typically in the several hundred hertz region, below which we simply cannot get good first-arrival-sound data. We can get good in-room data all the way down, but we can usually only get good first-arrival sound data down to several hundred hertz.

" However, from a perception standpoint, correcting on total volume is not a bad idea as you will have both speaker irregularities and room peaks and valleys that are not in the recording so removing them, on balance is better."

Imo the first-arrival sound is what matters most north of the Schroeder frequency region (yeah I know that term is debatable within the context of a small room). If the off-axis response (which dominates the "room response") has not been addressed by the loudspeaker design, then address it with acoustic treatment if we can. And if we can’t THEN imo it might well make sense to compromise the first-arrival sound for the sake of the overall tonal balance, but I’m not sure I’d trust an algorithm to make those decisions - I’d want to make the adjustments by hand.

Imo what we definitely DO NOT want is that a DIFFERENT "room EQ" be applied to the left and right channels (north of the Schroeder frequency), as despite the term what we’re REALLY doing is EQing the first-arrival sound, and we don’t want to introduce a channel mis-match into the all-important first-arrival sound. (Just to be clear imo the in-room sound matters a LOT, but the most effective place to address it is at the loudspeaker design stage, followed by room treatments.)

" I can’t see latency being an issue as it would be consistent across channels by design (or intentionally different)."

If we run everything through the DSP unit, that is true. If we only run the subwoofer’s signal through the DSP unit, then latency could be an issue. And not everyone wants to run the signal going to their mains through a DSP unit. So, imo, whether or not latency is an issue "depends on the specifics".

Duke