ETC Calculations

[mike w]

Hi Matt: I’m looking for help in understanding something about the ETC graph. See attached. The string test says I should be able to identify a point of reflection by the time of flight captured on the ETC graph. Example: The peak at the right speaker at 10.3 ms can be converted into distance (ft) by multiplying by 1.1 to yield ~11.3’. This distance would correspond to the floor bounce between me and the speaker. Ok.

If the direct distance between me and my speaker is 9.5’, how can I have anything less than 8.6ms show up on the ETC graph? How could something, no matter how near, reflect off something else and arrive before the direct sound? Do I misunderstand the time zero here? In which case, my calculation for the floor bounce and anything else would be wrong.

BTW, my speakers are set-up asymmetrically in an asymmetric room.

Oh, one more thing. If the level is down -20dB, does any of this matter?

Thanks,

Mike

 

[John Mulcahy]

The initial peak is the direct sound, so the delays to later peaks need to be added to whatever the direct path delay is.

20 dB down is good though.

 

[mike w]

Thanks John. Is there more to it than that? Please see response I read from Localhost127 in another forum where he is berating someone for summing the direct path distance with the indirect arrival time. I'm not sure if its about mixing units or something else. Localhost mentions the need for a loopback test. Do you know what he is referring to? He seems to be a very competent fellow (as well as entertaining).

"and nevermind the fact that a few weeks ago, kxxyyyzzp (who couldnt even convulve an ETC and was stuck with the 'IR' in REW) - also tried telling me that loopback is not required for identification of early arriving high gain indirect signals. he thought he could simply "add the direct path distance" to the indirect arrival time. lol. he doesn't even understand a simple measurement tool yet thinks he is authorized to participate in the discussion here. dont even get me started on his complete lack of understanding of porous absorption!" AVSForum 2012

My goal is to identify and treat early reflections.

regards,

mike

 

[John Mulcahy]

No, no more to it. The best tool for determining the direct path distance is a tape measure

 

[Matthew J Poes]

Hah I like Johns response. He is right.

 

[Matthew J Poes]

I’ll just add, using a single Omni mic to measure and identify reflections in a room is tricky. It can be done, but there are better ways.

The best way to think about it is to work out the geometry. What is the distance between the speaker and mic? Then work out the total distance between the speaker, a reflection point, and the mic. The difference in the distance between that and the first measurement is the distance to the reflecting surface. The problem is, many reflecting surfaces are often a similar distance so figuring out from the measurement exactly which bunch of noise in that measurement is coming from which reflection point can be tricky. The better approach is through the use of a sort of triangulation approach. Something like a 3D impulse response (REW can’t do this) or even just a set of carefully placed measurements with the distances hand measured to each other in the speaker. The change in the timing of the peaks tells you the location of the reflection in a 3D space.

None of this is worth the effort in my opinion. If you want to treat the first reflections, simply treat them. If you want to see the effect you can look at the difference in the ETC or possibly the change in the wavelet.

I also wouldn’t assume that treating first reflections is a good thing. Blind testing of the approach found people actually prefer the reflections intact. Floyd Toole basically says do what sounds best to you personally, but arguably leaving them in is better.

 

[mike w]

Thanks for weighing in Matt. I did identify some early reflections in the ETC graph by the string method. I put a sofa cushion at one FRP and confirmed it on the ETC. However, the distortion graph was horrible with the cushion in place.

Next, I added a temporary extension to my diffusor in that same space and again, the ETC looks better but the rest of the graphs (distortion, waterfall) do not. Moreover, I could not hear a difference one way or another. So, the do-nothing option won out.

This effort was driven by an old thread on Gearslutz where one participant argued (compellingly) that fixing things in the time domain with ETC cleans up things in the frequency domain. I did not find that to be the case. Maybe my late reflections are already below threshold (20 dB down).

I think I’m done optimizing. Now, if I can curb this compulsive behavior to tweak the acoustics, I could listen happily ever after.

 

[Matthew J Poes]

Thanks for weighing in Matt. I did identify some early reflections in the ETC graph by the string method. I put a sofa cushion at one FRP and confirmed it on the ETC. However, the distortion graph was horrible with the cushion in place.

Next, I added a temporary extension to my diffusor in that same space and again, the ETC looks better but the rest of the graphs (distortion, waterfall) do not. Moreover, I could not hear a difference one way or another. So, the do-nothing option won out.

This effort was driven by an old thread on Gearslutz where one participant argued (compellingly) that fixing things in the time domain with ETC cleans up things in the frequency domain. I did not find that to be the case. Maybe my late reflections are already below threshold (20 dB down).

I think I’m done optimizing. Now, if I can curb this compulsive behavior to tweak the acoustics, I could listen happily ever after.

Eh...I mean...if you remove all of the reflections the response improves, an anechoic response is much cleaner than an in-room response (But I think most would agree that isn't a desirable trait). However, getting rid of some reflections doesn't typically make a big noticeable difference in the steady-state response. you have to completely remove large quantities.

Just as an example, I find that acoustic ceilings make a noticable difference in the measurement response of a room. They all tend to measure similarly. That makes sense, the entire ceiling is highly absorptive at mid to low frequencies. Adding a couple bass traps to a ceiling tends to do close to nothing to the steady-state measurements.

Try to keep in mind, what matters isn't what the measurement looks like without some consideration for the psycho-acoustics. That is, just because it looks different doesn't mean it's better. I'd be careful chasing a specific ETC curve as there is no research that I'm aware of that shows the reduction of first reflections improves sound (this common convention was actually disproved was researched).

My own opinion is that we should use in room measurements primarily for the purpose of generally assessing the rooms acoustics and the general shape of the speaker response, to detect and eliminate bass modes (That is really it's main use for most), and that is about it. The rest should be based on what sounds good to us. You can drive yourself nuts chasing these measurement differences and never hear a wit of difference.

 

[thexder]

The initial peak is the direct sound, so the delays to later peaks need to be added to whatever the direct path delay is.

I'm looking at the REW 101 guide where ETC graph is showing us reflections. First one is marked at 5 ms.
Again, an amateur question.
I suppose the initial peak is at zero?
Or is the one at approx 1,5 ms showing the direct sound?

 

[John Mulcahy]

Initial peak = first peak, and except in very rare cases the largest peak.

 

[thexder]

So the one, marked with red circle needs no attention?

 

[John Mulcahy]

You have lost me, sorry, what does that have to do with your question about the initial peak? The circled peak is after the initial peak, it is a reflection.

 

[thexder]

The circled peak is after the initial peak, it is a reflection.

Ok, that was my question. Just wanted to be 100 % sure.
I calculated the distance of the first peak and it looks like it is very close to the speakers. No matter what I do it's always there. I have some effect on 2nd, 3rd etc ... but first one is always unchanged.
Time for rope and tape. Again.

 

[John Mulcahy]

Such reflections could also come from surfaces or features close to the mic, in case it isn't clear of walls, furniture etc.

 

[thexder]

Yes, it goes both ways. I tried to isolate the mic with some foam but first peak remained.