Need Room Treatment Advice. Willing to pay! - Page 3

Moderator warning

several reports received: please keep the discussion technical and avoid any bickering so we may all benefit from this thread

edit: my room is similar to OP so I am trying to get some free tips

We have probably given the OP more than he asked for. But I hope all of our contributions have widened the subject and given pause to those that feel they are only one step away from their ultimate goal. This thread is a good case example of how deep the well can get. The OP or anyone else reading this thread can decide for themselves at what layer of the Rabbit hole is sufficient for themselves to stop at. I never feel too much information is a bad thing. I don't think any of us can assume what piece of information here couldn't be relevant for someone.

Regardless of such minutiae, the main point in the context of this discussion is that delayed reflections create comb filtering. The more alike the two sound sources are, the worse the peaks and nulls will be.

--Ethan

What you really mean is the more exact the reflections are, the worse the peaks and nulls will be. Right?

Yes, one source is the direct sound and the other is the reflected sound. When the level of the reflections is the same as the direct sound, you get 6 dB peaks and infinitely deep nulls. Of course, reflections are never perfect, nor the same at all frequencies, and the volume is lower due to distance and also less than perfect reflectivity. But that's the basics of what causes comb filtering.

---Ethan

I was trying to say (badly) that with typical speaker / listener positioning for home theater and two channel, with speakers say 9ft+ away from the listener and in typically sized domestic rooms, the spectrum of early reflections dominates the frequency response at the listening position above the transition frequency. As contrasted to say 'near-field' listening like that used with two monitors on a bridge in a control room, where the direct sound dominates.

Even if you could hold perfectly still, without being able to "move your head slightly" (to use your description), both ears cannot be the same distance from both speakers simultaneously. So delayed copies of the same (mono) sound arrive at each ear. Based on what you said, each ear should be experiencing a comb filtered signal. Why haven't people complained about it over all the decades that 2-speaker playback has been popular?

Besides, people don't always sit precisely between two speakers, let alone sit perfectly still for long durations. When listeners sit off-centre, the common complaint is that the phantom centre pulls towards the side they're sitting, which is easy to demonstrate and hear. How come no similar complaints from off-centre listeners about hearing severe comb filtering when listening to a dual-mono source with different delays?

How can it be a problem for two similar sources (speaker and reflection) while simultaneously not being a problem for two similar sources (left and right speaker)?

^^^

no such thing as 'small signal' folk on AVS: just members who help each other

the OP was asking for some basic help: and it seems we have yet another thread cluttered with minutia that is interesting but not really helpful

Thanks to those experts who did try to help the OP : we need more of that in this thread

The fallacy I see above is the presumption that something can't exist unless people are complaining vociferously about it. ;-)

Comb filtering is found every time an audio signal is mixed with an unsynchronized copy of itself. Here is an example of comb filtering due to a 10 mSec delay:



The real world is full of reflections with one or more time delays that mix when we hear. It seems safe to say that human hearing evolved in such a way that we hear effectively in the presence of considerable amounts of comb filtering.

It is thus illogical to think that a listener would object to the presence of comb filtering unless there was an extreme case of it.

An associated question would be the nature of the perception of comb filtering.
Edited by arnyk - 1/17/13 at 8:48am

Quite often, a listener may have an objection to something they are hearing, but do not know the cause. Comb filtering is not a household term that the general layman is going to have a firm definition of or the recognition to know what it is when they hear it. In untreated or under treated small rooms, its likely comb filtering may be audible if not down right objectionable. But the listener will simply think their speakers or equipment doesnt sound good.

I had a comb filtering problem in the 3k-6k range that back in the day, I had no idea of the cause. I keep using a EQ to turn down those frequencies. While this helped, it also made my response quite uneven. It wasnt till sometime later after adding more treatment that the problem went away, and only then started to piece together what had transpired.

Another strawman. No one is questioning the actual existence of comb filtering, since that can be easily demonstrated through measurements. I'm asking why it hasn't been an audible problem historically, based on the description Ethan gave (more similar the two signals are, the worse the comb filtering).

With that description in mind, worst case scenario would be the same sound coming from 2 speakers, arriving at the listeners ears with a timing difference (walls aren't perfect mirror reflectors, so reflections might not be flawless copies of the direct sound).

Everyone has heard folks justify the use of a centre speaker to avoid a phantom centre image collapse to the nearest speaker for off-centre listeners. When was the last time you heard someone justify use of a centre speaker in order to avoid severe comb filtering when dual-mono dialogue is heard by off-centre listeners?

Would that be more audible with a single errant reflection in a heavily treated room or when you're in a sea of reflections in an untreated or under treated room?

I single errant reflection of the right magnitude, right angle (phase) to the direct signal could certainly do it. Even a well treated room only has to have one single reflection in the right direction and under the right conditions to cause comb filtering. So a well treated room in of itself doesn't guarantee you will avoid the problem. On the other hand, a untreated room could avoid the problem due to favorable geometry.

I am not an expert on this subject, so I cant even speculate on the character of multiple reflections and their effect. I believe that comb filtering effects can be seen on a FR graph at high res (like 1/48th smoothing or less). A very jagged series of peaks and nulls close together I believe is its signature.

Wisdom!

Honestly this kind of graph, which a lot of people post versions of when the subject of comb filtering comes up, is meaningless in the context of comb filtering at the listening position in a typical listening room / home theater because it does not take into account two important things:
1) fall off in SPL for the reflected sound, as it has a longer path to travel than the direct sound
2) fall off in SPL off axis due to speaker directivity

Ethan's comb filtering graph that he linked to earlier is also over stated because the mic and speaker were so close to the wall.

It could also reduce the audibility of the problem: with many reflections coming from many directions simultaneously, it is unlikely that the effects of a single reflection would attract your attention.

I don't think I posted a graph in this thread, but any graphs I've posted in other threads were real measurements. I do agree that a graph showing electrical comb filtering is overstated for the reasons you gave.

Attached are two graphs measured in the RealTraps lab room (16 x 11.5 x 8 feet) in a normal near field listening position, not near to a wall. One shows the range from 1 KHz to 10 KHz, and the other is just 2 KHZ to 4 KHz to better see the comb filtering detail. Again, this is a real measurement showing the sound that reaches your ears in a room with bare walls.

--Ethan

To make sure i am interpreting the graph right. Its the rapid succession of peaks and valleys closely together giving the indication of comb filtering, yes?

Do you have a before and after (treated vs not) of the same space illustrating both what a comb filtered response looks like vs one that doesnt? (a link would be fine for this also)

Those measurements are unsmoothed, so again misleading. You know as well as I do that we don't hear what the graph shows. And what did you mean by 'a normal nearfield listening position'? AFAIK our normal listening position is not what you would call nearfield which I interpret as where the contribution of the direct sound to measured frequency response is greater than the reflected sounds.

Why do you think comb filtering is psychoacoustically relevant? The only area I can see it being relevant is the first couple of peaks and dips in the gap between the room's modal region (where room modes are swamping the effects of comb filtering) and the region where our ear's loudness summing bands (ERBs) make any repetitive comb filtering irrelevant. But maybe I am missing something?!

I agree that comb filtering is more important in the case of a control room or other pro type environment where the speakers are very close to you.
Edited by Nyal Mellor - 1/18/13 at 12:54pm

Of course the plot above is idealized. I was trying to keep it simple.

However, I have measured deep nulls at regular intervals in real world listening rooms. I've encountered nulls that were so regular that the location of the next null could be predicted within a fraction of an inch after measuring a few others with a tape measure. In some cases the nulls are so deep that it sounded like the sound source completely went away if I blocked one ear and put the other ear at the null.

So, all those things you mention can presumably happen, and we still end up with pretty severe comb filtering.

BTW, if you can identify the surface that is causing the reflection that creates the nulls and treat it with appropriate absorbtion, the nulls can be vastly reduced, and in some cases largely eliminated.

No magic, just science. But you have to overcome denial... ;-)

Here is my ETC (200us smoothing) in 2 versions.

Red = My normal setup
Blue = Covering the panel responsible for my 24ms termination




Here is the FR with 1/12th smoothing from 1k-7k




Here is the FR with no smoothing from 1k - 7k



Thought the info might be insightful and/or worthy of analysis.

I'm not in denial Personally I just don't think it is that important. The first bounce null (aka speaker boundary interference) is the most destructive, after that it's importance goes down as the dips and peaks become smaller and our ear's loudness summing bandwidth becomes greater. And especially in the context of discussions about reflections where people post those unsmoothed (Ethan) or super simplified (yours) comb filter graphs and readers who don't understand the nuances immediately think 'oh wow I must absorb or diffuse all reflections' otherwise that nasty comb filter is going to mess up my sound! The reality is far from that.

Thanks for posting! Kind of funny (to me) but your ETC shows you have likely done a lot with absorbing, redirecting or diffusing your reflections yet if you look at the unsmoothed measurement which is like the one Ethan posted it is still looks like a comb filtered response!

You may be preaching to the choir. You can't expect everybody to summarize their complete philosophy of room acoustics in every post. For example, 2 weeks ago I wrote the following:

http://www.avsforum.com/t/1446589/need-some-help-for-living-room-acoustic-treatment/60#post_22781071

"Also, a good sounding room comes from the balance between reflection, diffusion, and absorption. I don't know of anybody whose preferred listening room is an anechoic chamber or a bare basketball court. Good is someplace in-between."

and that was not the first time and AVS was not the first place I posted similar words...

Well, I am trying to understand. But clearly, Ethan non-smoothed response is quite different than mine in the sense of the ranges from peaks to valleys. His range is about 20-30db (peak to valley), mine are about 10db with the 24ms peak return, about 5db without it.

But yes, even without the 24ms -10db return, there is a bit of a comb signature to it. I dont know how much is good or bad, and I am betting many other do not either. But if someone can use my illustration to make conclusions, please do, for good or bad in my case

I included the 1/12th response for your benefit really since you made the statement the the unsmoothed response isnt what the ear actually hears. In the 1/12th response, you can see a less than 1db difference caused by the -10db 24ms return. Only under close scrutiny (the unsmoothed response) can you see what that return is doing.


I can say from listening, and having an idea what comb filtering sounds like, that the -10db 24ms return doesnt sound combed. It sounds smooth and silky. So I guess the point here is to try to determine how much comb filtering need show in unsmoothed curves to be noticeable. If Ethan's bare room response DID sound combed, and mine does not, this may say something about what the peak to valley combing needs to be to be audible. In this case study, it would seem to be somewhere between 10db spreads (mine) and 30db ones (his).

Sorry if I was I don't post here regularly enough to know everyone's 'position' on acoustics