RoomFit is very good - my experience and settings

[dominikz]

Quick question: Should the microphone be pointed directly at each speaker, or kept in the sweet spot for both channel measurements?

If you only apply RoomFit at low frequencies (which I'd recommend) mic orientation is basically irrelevant.

If you'd like to EQ higher frequencies as well, or if you just want to be as precise as possible, I'd advise to put the mic in the sweetspot, point it up to the ceiling, and use the 90° mic calibration file. This way the mic response is identical (and correct) for sounds coming from all directions in the horizontal plane.

 

[harkpabst]

At least with the UMIK-1 and UMIK-2 the 90 deg calibration file introduces one more potential source of error, since it's just calculated, not measured.

After reading (I won't call it researching) many arguments of true experts (much more knowledgeable than me) I've settled for pointing the mic straight when doing swipes in 2 channel systems. Even with the moving microphone method (applied over a larger area) I'd prefer varying the microphone angle over vertical.

From my point of view the situation is quite a bit different from the typical multi-channel measuring procedure where most systems try to automatically determine the distance to each speaker from one MLP.

But as always, YMMV.

 

[dominikz]

At least with the UMIK-1 and UMIK-2 the 90 deg calibration file introduces one more potential source of error, since it's just calculated, not measured.

After reading (I won't call it researching) many arguments of true experts (much more knowledgeable than me) I've settled for pointing the mic straight when doing swipes in 2 channel systems. Even with the moving microphone method (applied over a larger area) I'd prefer varying the microphone angle over vertical.

From my point of view the situation is quite a bit different from the typical multi-channel measuring procedure where most systems try to automatically determine the distance to each speaker from one MLP.

But as always, YMMV.

Let me start by repeating that for the purpose of room correction at low frequencies measurement microphone orientation is IMHO irrelevant, so any approach works just as well.

This means that this discussion is largely academic - but perhaps still interesting to some!

Let's first remember that measurement microphones are omnidirectional at low frequencies but become more directional at higher frequencies.

Therefore:

• If we use the 90° calibration curve the mic will capture any sound coming from its sides with a flat spectrum, sound coming at the front will have some HF boost, and sound coming from behind the mic will have some HF attenuation (compared to flat). So if you point the microphone vertically to the ceiling, direct sound, side-, front- and back-wall reflections will all be registered with the correct spectrum by the microphone as the will all hit it at the sides! Ceiling reflection will have some HF boost coming at the front of the mic, and floor reflection will have some HF drop coming from its back, but this is a smaller part of the total captured sound so the total balance of the in-room response should be roughly correct.
• If we use the 0° calibration curve only the sound coming from the front of the mic will be registered with a fully flat spectrum, while sounds coming to the mic from any other direction will exhibit some amount of HF attenuation. So if we point the microphone horizontally between the speakers, most of the horizontal or vertical reflections will be captured with some HF roll-off, resulting in some (relatively minor) HF loss in the total measured in-room response.

This is something I noticed experimentally as well - when doing MMM with vertical mic orientation and 90° cal file I get a slightly higher response at the highest frequencies, compared to pointing the mic horizontally and using a 0° cal file.

In addition, IME MMM measurements with mic in vertical position and 90° cal file track the anechoic PIR very closely at high frequencies, here's one example:

 

[harkpabst]

This degree of correlation between MMM and Klippel PIR is surprisingly good, indeed.

Did you try and compare it to a 0⁰ measurement? Either strictly horizontal or pointing at different angles, as I lightheartedly proposed?

 

[dominikz]

This degree of correlation between MMM and Klippel PIR is surprisingly good, indeed.

Did you try and compare it to a 0⁰ measurement? Either strictly horizontal or pointing at different angles, as I lightheartedly proposed?

Not sure TBH, as it was a while ago since I looked into this.
There's some potentially interesting (though inconclusive) examples in this post, but it's not MMM and there's no comparison with PIR.
Perhaps I'll test this again when I find some time.

 

[Mr Ee]

I don't quite understand the difference.

"Should the microphone be pointed directly at each speaker, or kept in the sweet spot"

Say you are equilateral to each speaker, I'm pretty sure it means should the mic be pointed 30° left then 30° right to point at each speaker or just pointing straight ahead.
Maybe immaterial for those with fancy omnidirectional microphones