Incorporating environmental factors for DRC

[3ll3d00d]

However, I do wonder if I am over-thinking it

I don't think it seems like overthinking it, more like conflating different concerns. I'll leave the HRTF part to one side as I have no idea about that aspect. The other part (furniture etc position) seems to go back to what are you trying to correct? if you think that the point of higher frequency correction is to address the speaker then ideally you'll be able to accurately measure the speaker itself. If you can only measure it in room with the speaker installed as normal then you have variety of effects to contend with which include those attributable to the microphone and nearby reflections (eg the seat).

If you intend to correct the speaker but can't measure the speaker itself then for sure it makes sense to move the furniture.
If you believe you should also correct for the local environment then you shouldn't.

 

[kemmler3D]

Since I don't aim to correct the speaker so much as in-room response, I do MMM at the listening position(s) and confirm with SPS, but I leave all the furniture in place. If you are correcting the room, isn't the furniture part of the room?

 

[Keith_W]

I don't think it seems like overthinking it, more like conflating different concerns. I'll leave the HRTF part to one side as I have no idea about that aspect. The other part (furniture etc position) seems to go back to what are you trying to correct? if you think that the point of higher frequency correction is to address the speaker then ideally you'll be able to accurately measure the speaker itself. If you can only measure it in room with the speaker installed as normal then you have variety of effects to contend with which include those attributable to the microphone and nearby reflections (eg the seat).

If you intend to correct the speaker but can't measure the speaker itself then for sure it makes sense to move the furniture.
If you believe you should also correct for the local environment then you shouldn't.

The purpose of moving the furniture is not for speaker correction. It is for room correction. These are 2 different things. I do the speaker correction first, and then I do the room correction after that. In other words, mic nearfield first, and then mic at MLP.

There are 2 types of "speaker correction" and it depends on whether you have your crossover separated and controlled by DSP or not. If your XO is separated, you can do driver correction. The mic is almost touching the driver when the sweep is taken, so concerns about moving furniture are moot. If the XO is not separated, the entire speaker needs to be corrected with a nearfield measurement or quasi-anechoic measurement, and the mic is placed about 1.5x the distance between the furthest drivers.

I know that Audiolense has a different workflow for active speakers. You place the mic at the MLP and then it does individual driver sweeps. I can see why they recommend moving furniture in that case. If I wanted to, I could do that with Acourate as well, but then it would be more difficult to separate the speaker response from the room response. This is why I don't perform driver correction with the mic at the MLP.

 

[AudioJester]

Heres w a weird experiment I tried a while ago.

The premise being a perfect measured frequency response does not account for my own hearing.

Using a hearing test (not great resolution) i created my own receptive hearing response curve. That is hearing sensitivity at different frequencies for each ear. (I am fortunate that overall my hearing is pretty good for my age, but there are no ideal curves). Then overlay this inverse of this curve to your dsp.

Now the room and your hearing is out of the equation.
If you can try it, curious what you think.

Oh, and this is not proper science more a fun experiment

 

[3ll3d00d]

The purpose of moving the furniture is not for speaker correction. It is for room correction. These are 2 different things. I do the speaker correction first, and then I do the room correction after that. In other words, mic nearfield first, and then mic at MLP.

I think you're missing my point

speaker/driver correction requires a measurement of the speaker itself which, for most people and most speakers, means some form of quasi anechoic measurement. Quasi anechoic measurements are taken by capturing the speaker with as long a reflection free zone as you can possibly achieve and then place the window at that point. "As long a reflection free zone as possible" means remove all obstacles that sit between the speaker and the mic in all directions. Doing this inside is possible and usually means move the speaker somewhere into the middle of the room & get furniture etc out of the way. At worst, you should end up with data good from ~1kHz but you can certainly get lower if you have a bigger space (e.g. massive room, outside with the speaker high in the air)

Compare this to measuring a speaker in a room with an fdw applied, now you capture whatever is in the way (walls, furniture etc) in the measurement but attempt to eliminate the effect of those reflections by the use of a window that gets progressively shorter as you go higher in frequency, i.e. at higher frequencies it's attempting to achieve exactly the same thing as a quasi anechoic measurement (measure the speaker itself).The more things you move out of the way when taking such a measurement, the closer you get to just measuring the speaker itself and the less it looks like room correction.

At which point, one can ask whether it makes sense to essentially try to correct the speaker twice and/or consider what exactly the room is contributing to those measurements (e.g. if you've used a longer fdw) and whether that is something you can/should actually correct.

If your XO is separated, you can do driver correction. The mic is almost touching the driver when the sweep is taken, so concerns about moving furniture are moot.

It's just not right to think that a NF measurement is good for speaker correction (except for a sub or woofer though then it's pointless as the room dominates and the driver is unlikely to need correction at all) given that the frequency limit for such measurements is 10950/D (where D is the effective diameter of the driver in cm). In either case (individual drivers or whole speaker), quasi anechoic data is what you really want.

 

[ernestcarl]

mic is almost touching the driver when the sweep is taken,

This first step considered as individual “driver” correction level seems questionable to me. It can be a useful data point, but I would not bother to make a correction filter from such a close proximity measurement. I know you eventually convolve multiple filters so the final filter ends up very different anyway — but, I think this is just one more way one ends up overcorrecting.

 

[goat76]

The best-sounding result I've had for adjusting the low-end frequency response is using MMM. I didn't sit on my sofa while doing the measurement, just standing behind the sofa and moving the microphone around the main listening position until the measured frequency response was stabilized. The result sounds better than I've ever achieved using fixed-point measurements. There are just three reducing adjustments, one large cut at 37 Hz and two smaller ones at 118 Hz and 359 Hz, other than that I'm satisfied with the overall sound. I have tried adjusting some of the deviations in the higher frequency area based on gated measurements but prefer the untouched response of my speakers for that area.

At some point, I will try making a new MMM measurement while sitting on the sofa and see if I can improve the bass response even further.