Thanks. So I did bump the curve slightly.
Here’s the After:
Here’s the “before” group delay:
The group delay looks pretty normal. The ramp up on the left is, I think, because the system works harder / room fights you - most systems show this. The spike at ~128 is probably a room mode. The magnitude plot looks ok though, so maybe not something to worry about. Moving the sub/main crossover might have an effect.
I usually get “after” measurements closer to the prediction. The trick is to try and use exactly the same mic locations. Regardless, if your ears are happy, then job done 
@gregwilding
Hello Greg,
Ok I’ve been doing some reading In future updates, do you think it’s possible/beneficial to include an option to input the passive crossover freq and slopes so the FIR can be optimized for that specific setup? I have no idea how hard that is or if the results would be worth the trouble.
What would you suggest the most beneficial freq. range is where any type of filters should be applied anyhow? I assume it’s a balance act not to eq too much?
thank you
Hi Cenk, neat idea.
As for frequency ranges / filter type - it’s more about knowing what’s worth correcting. We can hear the peaks and some dips at low frequencies (below ~300 Hz), but as frequency increases, they get hard to hear and aren’t worth fixing. Thus, for room correction, we want a filter to pay more attention to the low end. At higher frequencies, the filter is just shaping the system response to a desired target curve. HouseCurve applies this logic when creating PEQ or FIR filters, so either type can produce good results with a full range correction.
Hi Greg,
One thing I noticed after having applied FIR to sub 300hz, the treble seems to be much more pronounced. I guess my ears/brain have a much better sight of the highs once getting rid of the low end “mush”. In my case I could just attenuate the treble on the speakers. Granted I like low treble in general and my room is bright, Is there a way to apply correction to a given bottom and high region, leaving mids alone?
I can’t believe what a difference FIR made over PEQ, thank you so much for keeping at it with the app.
Cenk
Hello,
I’ve come to the realization that while correcting for a specific listening seat, the rest of the room can actually sound worse than before. Just wanted to share. There really is no free lunch in audio:(
I did try widening the mic distances, increasing the measurement count and limiting eq to 300hz. Seems to help.
Next trial will be doing the same trying moving mic with the short pink noise.
Hope it helps somebody.
Cenk
Yep. The wider the listening window needs to be, the more a departure from DSP for traditional stereo playback the objective becomes. It’s not an imperially-validated trend so far as I know, but it makes sense to hear about it each time the issue crops up.
Felt like my overclocking CPU teen days where squeezing the last drop of performance resulted in crashing the PC. I need to tweak less, an listen to music more.
@gregwilding
Hi Greg,
Why is the improvement delta so much more between the original and filtered signal when I upsample?
Whereas I don’t hear a difference when I compare native and upsampled non-filtered signal?
Thank you
Hey Cenk, I’m not sure I understand the question - what do you mean by improvement? and what filter type (FIR / PEQ)? I don’t think upsampling would do anything to the sound…
Hi Greg,
I used FIR and exported in multiple sampling rates, putting Camilla in variable sample rate mode. When I activate the upsampler of Camilla I get a higher clarity. Which I don’t when I do the same without a FIR filter applied on the signal.
So I was wondering if higher sampled FIR has more fine control the signal.
Thank you
Ok, I see. There’s nothing special about the higher rate filter. HouseCurve is just resampling for the different rates. My guess is that Camilla is doing something special when there’s a filter in the signal chain. Hehe, I guess if it pleases your ears, you’re winning 
I guess so. Yes it really makes a difference.
Hello Greg,
I realised if I keep the mic position EXACTLY the same for separate L and R channel sweeps, the woww factor is enormous for the sweet spot. With FIR that is. Is there a simpler way I can do sweeps L then R → move mic → L then R → move mic etc. Right now I have to go change saved measurements, switch to correct channel, switch sweep channel etc. Prone to mixups.
I hope you can duhhhh me on this with a simple trick
Hello, does anyone know if there’s a similar app on Android?
google on Room Acoustics Meter Android. Not going to place a link as I don’t want to hijack Greg’s topic.
There isn’t a simpler way at the moment. I mark the floor with tape to keep my measurement locations roughly the same. Then I measure a channel, save the average, etc. It’s a bit of a pain.
I’ve been considering a mode that measures channels separately and manages them as a set, exactly as you describe. It would take a fair amount of work to implement though.
An alternative is to make the app aware of the location a measurement was taken (ex: far left, left, center, right, far right). With that info it could synthesize an average left / right measurement from the data provided. That would be faster and may produce better averages. This could sort-of be done manually, ie: with both channels active, measure an average of left to center and and an average of right to center. I’ve been meaning to try this, but haven’t had the time.
Hi Greg,
What you describe in your last paragraph could even take care of cancellations between two speakers which separate measurements would not. At the cost of mashing the phases into one average I’d guess? Thanks for the tip.
maybe it’s easier to implement measurement import? than on could use two instances of HouseCurve on iphone and ipad for each channel and just replug the mic at each position? 
@gregwilding Hey Greg, Andrew Robinson mentioned HouseCurve in a video. yuhuuu🤸🏻
