Update.
RED speaker positioning has been measured, and adjusted and measured again, and listened - and repeated a few more times.
Not surprisingly the pre-room correction measurements were reasonable, but had a couple of deep nulls, right in the heart of much music.
View attachment 83012
View attachment 83013
View attachment 83014
So I moved speakers around, and changed the LP accordingly until I found a reasonable compromise. Working out where the mode is generated is helpful - width or height mode for example. My method is to simply play a tone at the problem frequency and move the mic around to determine where the nulls are and where the peaks are, seems like a fast way to avoid moving speakers more than required.
The effort of moving things a hundred millimeters or so gave the following result - generally much better through the mid range.
View attachment 83016
View attachment 83009
Then I ran DIRAC 3.0 on my miniDSP, and re-measured the results, adjusting the correction to get the best compromise - in the end not much measurable difference, but audibly much better mainly due to the impulse correction I expect. Difference with and without DSP is subtle.
View attachment 83024
View attachment 83027
View attachment 83026
View attachment 83029
View attachment 83028
Next steps are to measure the outdoors speaker position, and perhaps get a better idea how good it can be. Then work backwards. The DSP is adding clarity and fixing impulse and timing issues L to R but it also seems to add a layer of audible dullness to the sound. The clarity with the DAC only (playing one channel only) is far greater. So something going on with the combined sound of both channels.
Comments and suggestions are welcomed.
var smoothing is the best to analyse.
not sure how to interpret your data. would be intresting to find out what is causing the remaining cancellation. there should not be much going on regarding standing waves between the walls in you room. Maybe between floor and ceiling? https://mehlau.net/audio/floorbounce/
have you found out the frequencies which make all the glass vibrate?
also for you comined sound issue it would be intresting to see a chart overlay of left and right speaker
Mic calibration file loaded? 90° (e.g. pointing at ceiling) or 0° ?and the treble roll off is pretty sharp after 10kHz
The big null mode is a height mode. Sadly impossible to correct without subs I think.
Interested to see if I can remove the 300-400 Hz null though, and the treble roll off is pretty sharp after 10kHz, old drivers / crossover caps.
View attachment 83058View attachment 83059
And yes, great idea, I will investigate where the glass is resonant - have not experienced too much brightness, probably due to the HF roll off mentioned.
And yes next time I am measuring I will look at left and right - although I did measure the initial position - see below.
View attachment 83060
Spectrogram should be limited to 200-250 Hz. Use ETC to find reflections.
Mic calibration file loaded? 90° (e.g. pointing at ceiling) or 0° ?
Spectograms reproduced with better detail.
Before with speakers initially positioned deeper into the vee of the room.
View attachment 83133View attachment 83134
View attachment 83135
Speakers then moved out of the vee further, and when measurements improved, room correction with DIRAC was performed. Then I simply tailor the HF and LF portions of the correction to match the speakers natural curves.
View attachment 83136
This final version - setup 2 - is quite good. Still a small hiccup between 300 and 400 Hz, but the decay times seem much more linear from low to mid frequencies, compared to the initial DAC only measurements.
ETC - what should I be looking for - graphs are quite different? Pre-ringing on the DSP version is very measurable if not audible.
View attachment 83138View attachment 83139
someting is wrong with the ETC graphs. can you atach the mdat?
ETC: (RFZ is no way near )
View attachment 83210
Ok, what is RFZ and what is nowhere near? I am new to all this and definitely still learning.
Spectrogram, normaliced:
View attachment 83211
What setting are you using - not wavelet?
Or use waterfall diagram, 60 dB range (you should measure at least 10 dB louder)
View attachment 83212
T60M/T30 is high, way to high.
Ok, so measurements should be peaking at 95dB? And what settings do you have the waterfall diagram on? The valleys look different.
Please add pics of your current setup. I'm suprised ETC is that bad. Bass not EQ, no spurise here.
damn... Just seeing the skyline of the CBD gives me thrills and makes me emotional. I miss my Aussie friends and living there so much. Best city in the world. Enjoy your new place ! looks like Bondi Junction, the view is fantastic.
ETC:
You have a high level for early reflections (-5dB relative to the direct sound), and a very long time for sound to decay (about 600ms)
View attachment 83318
(An example of mine, for comparison - first reflections about -25dB and 300ms settling time)
View attachment 83321
The impulse response shows the time the reflections occur relative to the direct sound at 0ms:
View attachment 83319
(an example of mine for comparison - reflections low in relative level and few in number)
View attachment 83320
So...
Consider my examples something of an extreme - smaller open furnished room, MartinLogan dipoles, and a little deadening beyond the couch and carpet.
You have a nice view, the sound is probably good at low to moderate levels, might sound like an echo box at high levels.
Imaging - I don't know, might be acceptable, might not, depending on what you like.
After the initial reflections, it would seem to be much more diffused (probably good).
There are suggestions to put it all in a closed room.
I think I'd live with the view and work on it a little.
The reflections surely are the floor and windows and probably the ceiling and bounces from one "wall" to the other.
My supposition would be that the curtains shown do next to nothing sonically.
Move them as close as possible to the wall and angle them in. Maybe even more than 30°. Absorb ceiling reflections.So what do you think?
Bass response can be measured nearfield (mic very close to center of woofer). Speaker itself with mic at 1 m, absorn floor bounce and gate signal to eliminate reflections (5 ms would be good). Try setting a gate in REW and see what it does.Any tips for measuring outside?
Thoughts, suggestions?