This is a post I made elsewhere to provide a high-level but detailed overview of how to look at room acoustics. It is the heart of the bible I follow. Hope you find it useful.
----
At high level, the key thing that unlocked this mystery for me, and hopefully will do for you, is to understand some fundamental points about sound reproduction in our homes. Once you have that knowledge, it will then enable you to easily understand people’s argument and a lens through which to evaluate them. Without it, you will be flailing in the wind. Here they are:
Here is a sample full frequency measurement:
What do we see? At 60 Hz the response peaks to 76 dB. At 20 KHz, the response is down to 37 dB. Put another way, his 20 to 20Khz frequency response varies by a whopping 40 dB!!! What happened to +- 3db that we often hear? Now maybe some of that is due to measurement errors. But even if I stop at 5 KHz, he has 20 dB fluctuations. That is hugely audible.
Second, we see the wild fluctuations in low frequencies as with the measurement I post at the beginning of this thread. Just the same, his high frequencies are much smoother. If you read the thread, you see that most of the discussion is around acoustic treatments. Well, he tried that. Here are his results with or without so called corner bass traps he built out of triangle fiberglass pieces:
Translation: it didn’t make a darn bit of difference! Now you see why my low frequency optimization article does not even talk about using acoustic products. While they can be useful (especially in the hands of a pro), far more powerful tools are at our disposal. In this case OP has 4 subs already. So while better placement could be had than lining them up on the front wall as he has done, I am not going to suggest that he put more in there as he might throw a rock at me
. We can however use one of our other tools: EQ. If we had a parametric EQ, we could set it to the center frequencies of those peaks, adjust the bandwidth to match it, and then pull down that level. He has one of these for example a hair below 80 Hz.
If we did that we not only would reduce the boominess he is getting from that, but increased clarity and definition. This is because when we fix such resonances in frequency domain, transients in time domain take less time to settle down. Popular music has a rhythm of about 0.4 seconds. Room resonances cause reverberations that are longer than that, causing distinct bass notes to run into each other. Pull that peak down and you not only get smoother frequency response but also better definition.
I should note that you will lose some of your bass power. So there is a cost to it and you have to compensate with more powerful subs.
Now read the thread and tell me how much of this simple insight was shared. You will see very little of it. Instead, he is told to keep messing with acoustic products and pick up woodworking as another hobby to build them! As a fellow woodworker, I don’t mind that advice but it really is not the right solution. He can pick up a cheap DSP box and put it in line with his subs and go to town there. If you are buying new subs yourself, look for ones with DSP built in. Even a single band EQ is of great help but may provide three of them which is quite sufficient to tame the low frequencies.
Throughout this thread and elsewhere, vocal opposition keeps talking about “specular reflections” and how you need ETC type analysis to find and fix them. Well, OP in that thread ran such an analysis. Here it is:
I ask you: what on earth does this tell you? Compare it to the frequency analysis I discussed above. Which is telling you the story better and what you need to do immediately to fix your room issues? Clearly it is the frequency response chart. So take Dr. Toole’s advice and ignore these charts and techniques that are devoid of the frequency spectrums. Your ears can easily tell if you screw up the frequency response. Time domain however is subject of psychoacoustics and not so easily analyzed.
Well, here it is. Now you have some of the high level points, I suggest buying Dr. Toole’s book and digging deep. It will all make much more sense now than if you had read it cold. And oh, read it multiple times. The more you learn, the more you pick up on the finer points.
----
At high level, the key thing that unlocked this mystery for me, and hopefully will do for you, is to understand some fundamental points about sound reproduction in our homes. Once you have that knowledge, it will then enable you to easily understand people’s argument and a lens through which to evaluate them. Without it, you will be flailing in the wind. Here they are:
- Sound behaves differently in your room depending on its frequency. Here is the opening graph and point in my article on low frequency optimization in your room:
As you see there is a critical band called the transition frequency. I won’t get into how that is calculated but for now, assume it is in the 200 to 400 Hz range depending on the size and attributes of your room. Below that we have distinct patterns of reflections interacting with each other, creating very significant frequency response anomalies. Above that, the reflections become chaotic and how they interact with each other highly complex, resulting in very different impact on the frequency response of the speaker. In the middle, during the transition frequencies, a bit of both is going on.
- Now look at the notations on the graph. To the left of transition frequency it says the room has the largest impact. Indeed it does. Massive frequency response variations exist due to so called “room modes.” You can change your speaker all you want and you will still get massive peaks and valleys as shown. Therefore, if there is one area that needs immediate attention, it is in that region! This is why the first installment of my articles on room acoustics started there. Get this region right and you are well on your way to good sound.
- Below transition frequency, measurements tell the whole story. Psychoacoustics does not enter the equation. You see a 20 dB peak at 50 Hz (for example)? You have a problem. There is no way that is not audible. And frequency response measurements tell you the exact problem you have. This is a great thing as we all understand frequency response graphs. It is the bible of high fidelity in audio. You want a flat response there.
- As frequencies get lower, the effectiveness of acoustic products such as absorbers and diffusers goes down. So with some rare exceptions (such as tuned absorbers), you simply cannot think of solving those problems with putting acoustic material on the wall. Part of the solution is to have flexible walls that absorb energy. But much more powerful techniques exists with making the problem less severe by better placement of speakers/listeners, and deploying multiple subwoofers and electronic correction (both EQ and Sound Field Management). A detailed and high level view of that is covered in my article: http://www.madronadigital.com/Library/BassOptimization.html
Note that if you have not done anything about your room’s low frequency response, then you are guaranteed to have problems here. Get a mic and run a frequency sweep and be ready to be horrified!
So summarizing where we are so far, you need to optimize the low frequencies in your room and that optimization does not involve arguing about diffusion, vs. absorption, vs. first reflection, etc. All of that is for the next section, above the transition frequency.
- So now we get to where the debate is: what happens above transition frequency. As you have seen, you can have 1000 page debates about this. But most of that is for not! Why? Look at the chart again: it says that the speaker is in control there. If you don’t believe me, take two different speakers in your room. Do they sound different? Answer is obviously yes. So if you want good sound there, there is no getting around having a good speaker. Don’t think you can start with a bad speaker and then by magic, there is some acoustic product that goes on the wall that fixes such things.
- The sound that reaches you is a combination of reflections in the room and that of the direct sound of the speaker. Measuring the speaker in an acoustic chamber at one point in front gives the direct sound performance. If you augment that with many more measurements on horizontal and vertical axis, another view appears: what the reflections would sound like. After all, reflections are mostly made up of sounds going at a different angle than straight on sound of the speaker.
- Turns out you can predict with high confidence what a speaker would sound like above transition frequency with the above measurements! Let me repeat: you can mostly predict what a speaker will sound like in real rooms from measurements you perform in an acoustic chamber. Indeed that is what Harman has demonstrated by measuring a large sampling of speakers and showing how their in-room response can be computed from a 70-point measurement in the acoustic chamber. The correlation is something like 90%.
So if you are a capable designer, you can design a speaker that behaves and works very well in a real room. And you can design what the sound should be like to be “accurate.” Accuracy there means a smooth frequency response as the angle of sound changes from dead on. Poorly performing speakers have wild response as the angle changes. Those poor reflections combine with direct sound and color the overall sound you hear. It has been said that this was the origins of the so called LEDE or Live-end, Dead-End room model and the strong emphasis to do away with strong reflections. They did have a problem to fix, but instead of fixing the problem (speaker), they thought it was the room that needed fixing. Thankfully we have come a long way in the last three decades when these models were developed and know how to build much better speakers that are no broken this way.
So take a listen to your speaker. Do you like how it sounds in the mid to high frequencies? If not, it is time for a new speaker. Don’t go putting things on the wall hoping that is the path to better sound. Likely it is not and will cost you money fixing something that is better fixed in the source: the speaker.
- This is another key point: above transition frequencies psychoacoustics enters the equation and there is no getting around it. Psychoacoustics says that what you measure is not what you are potentially hearing. If I take a 128 Kbps compressed music and run a measurement on it, I might find horrendous distortion figures at times. Yet, to listeners it is often similar sound to the CD. How often do you see various encoders compared using harmonic distortion (THD) figures? Never. What people do instead is perform listening tests.
The same is true here. We have two ears and a brain that is interpreting what is almost always two different signals. When the speaker sound goes sideways hits the refection on the right wall and comes to you, you hear two versions of it. The right ear is closer to it so it hears it earlier. The left ear hears is 0.4 msec later and with its frequency response changed due to your head masking some of the sound. The brain then attempts to interpret what it is hearing. You would go mad if that were two different sounds. Fortunately adaptation kicks in because we hear reflections all the time and we *interpret* what we are hearing. We do not at all see a single measurement as people post in these discussions.
So run, and run fast if all someone wants to talk about is showing you measurements above transition frequencies. Listening test results must accompany any such discussion. And then back correlation to why we perceive what we perceive.
How many singers do you know that like to sing outside where there are almost no reflections? Have you tried to sing yourself there? Fact is that some of these effects are totally non-intuitive. We think a reflection is a bad thing. But clearly no singer thinks that. I know it is hard to let go of these notions but let go we must
- As we have heavily discussed, there are arguments over reflections and tendency to want to follow the “pros” as far as mixing and recording rooms. Having worked for companies that sold products to this industry, I am telling you, they have different reasons and priorities than you do enjoying movie and music at home. A post would not be complete without a quote from Dr. Toole . So here it is from his CEDIA Course:
” Musicians who perform live tend to be hypersensitive to reflected sounds –
judging them to be as much as 7 times as large as ordinary listeners. Why?
Perhaps because in concert venues what they hear of themselves playing and of
their colleagues playing is substantially learned from reflected sounds coming
back to them from the audience area and from the stage house. Recording
engineers are unusually critical of early reflections in recording control rooms.
Why? Perhaps because adding and manipulating reflected sounds are
fundamental operations in the creation of a mix. They can turn reflections on or
off, up or down, and thereby learn to distinguish components of sound that they
are creating from those existing in the room. Not surprisingly, they want the room
sounds to go away. Interestingly enough, there is some evidence that these
same people like lateral reflections when listening for entertainment at home.
But neither of these is home theater, and these people are not our average
customers. When put to the test, most listeners prefer additional reflected
sounds in small rooms. Alarmist talk about reducing the precision of image
localizations have not been proved. The much touted “comb filtering” turns out
not to be an audible problem, unless it is in the program itself, which is certainly
not our problem. On the positive side, it turns out that reflections increase our
ability to hear timbral subtleties in the program.
About the only negative that can be raised against early lateral reflections is that
they will mercilessly reveal loudspeakers that have poor off-axis frequency
response. Over the years there have been lots of these, including some prime
examples in the studio monitor category. The best solution in these instances is
to absorb the off-axis misbehavior. This, in fact, became a bit of a fashion in the
“dead end” control rooms of a couple of decades ago. But, why start with bad
loudspeakers to begin with?”
- The next step is to determine how reflective our room is. While some reflections are good, there comes a limit after which intelligibility suffers. Fortunately, most well furnished rooms are where they need to be. I described the formal test for this which is an RT60 measurement around 500 Hz and getting a value of 0.4 to 0.6 sec range. But an approximation is to have someone stand by your speakers and talk to you in normal voice. Are you able to understand them? If so, you are probably good to go. If you have trouble, it is likely too live. The reason is that the human voice has slightly wider dispersion than a typical speaker. So if it sounds good, then a speaker is going to sound good.
An empty dedicated room is likely to need help in this regard. A well-furnished living room however is probably fine based on large scale research into this area (some 600 homes were surveyed). The reason is that the typical furnishings and items in a room such as carpets, bookshelves, furniture, drapes, etc. all contribute to bringing the reverberation time down to a comfortable and desirable level. Note that we are not trying to achieve precision here but to get us in the ballpark.
- In a dedicated space the carpet can be an important element of bringing our reverberation time down. Ideally you would want to make that broadband by using the appropriate lay-up such as felt underlayment and the right carpet material. If we do the math based on the RT60 reverberation time mentioned before, you would need to cover 25%-30% of the rest of the surfaces with absorption material.
Such material should be broadband going all the way down to transition frequencies. Otherwise, you would be shaping the response of a good speaker by taking out its high frequency reflections and not low frequencies. So this usually calls for 4 inches of standard absorbers.
- Location-wise, we want to put absorbers where they will be eliminating reflections we don’t care about. A good choice here is front and back walls. The sound coming out the back of your speakers is going to have pretty poor response so doing away with it is a good thing. But importantly, we tend to prefer lateral, side-to-side reflections because that gives us more realism. Front to back reflections don’t accomplish this because the much stronger direct sound from the speaker masks these reflections. So they become a good spot for absorbers.
Additional ones can go on the side walls as needed, avoiding the lateral reflection points which you may want to enhance with diffusers to increase that realism.
Here is a sample full frequency measurement:
What do we see? At 60 Hz the response peaks to 76 dB. At 20 KHz, the response is down to 37 dB. Put another way, his 20 to 20Khz frequency response varies by a whopping 40 dB!!! What happened to +- 3db that we often hear? Now maybe some of that is due to measurement errors. But even if I stop at 5 KHz, he has 20 dB fluctuations. That is hugely audible.
Second, we see the wild fluctuations in low frequencies as with the measurement I post at the beginning of this thread. Just the same, his high frequencies are much smoother. If you read the thread, you see that most of the discussion is around acoustic treatments. Well, he tried that. Here are his results with or without so called corner bass traps he built out of triangle fiberglass pieces:
Translation: it didn’t make a darn bit of difference! Now you see why my low frequency optimization article does not even talk about using acoustic products. While they can be useful (especially in the hands of a pro), far more powerful tools are at our disposal. In this case OP has 4 subs already. So while better placement could be had than lining them up on the front wall as he has done, I am not going to suggest that he put more in there as he might throw a rock at me
. We can however use one of our other tools: EQ. If we had a parametric EQ, we could set it to the center frequencies of those peaks, adjust the bandwidth to match it, and then pull down that level. He has one of these for example a hair below 80 Hz.If we did that we not only would reduce the boominess he is getting from that, but increased clarity and definition. This is because when we fix such resonances in frequency domain, transients in time domain take less time to settle down. Popular music has a rhythm of about 0.4 seconds. Room resonances cause reverberations that are longer than that, causing distinct bass notes to run into each other. Pull that peak down and you not only get smoother frequency response but also better definition.
I should note that you will lose some of your bass power. So there is a cost to it and you have to compensate with more powerful subs.
Now read the thread and tell me how much of this simple insight was shared. You will see very little of it. Instead, he is told to keep messing with acoustic products and pick up woodworking as another hobby to build them! As a fellow woodworker, I don’t mind that advice but it really is not the right solution. He can pick up a cheap DSP box and put it in line with his subs and go to town there. If you are buying new subs yourself, look for ones with DSP built in. Even a single band EQ is of great help but may provide three of them which is quite sufficient to tame the low frequencies.
Throughout this thread and elsewhere, vocal opposition keeps talking about “specular reflections” and how you need ETC type analysis to find and fix them. Well, OP in that thread ran such an analysis. Here it is:
I ask you: what on earth does this tell you? Compare it to the frequency analysis I discussed above. Which is telling you the story better and what you need to do immediately to fix your room issues? Clearly it is the frequency response chart. So take Dr. Toole’s advice and ignore these charts and techniques that are devoid of the frequency spectrums. Your ears can easily tell if you screw up the frequency response. Time domain however is subject of psychoacoustics and not so easily analyzed.
Well, here it is
. Now you have some of the high level points, I suggest buying Dr. Toole’s book and digging deep. It will all make much more sense now than if you had read it cold. And oh, read it multiple times. The more you learn, the more you pick up on the finer points.
Last edited:
