How to Treat Small Room Acoustics

[gene]

For a long time, conventional wisdom has treated large and small room acoustics as one in the same. But there are fundamental differences and objectives between the two and thus they should be handled differently accordingly.

Small rooms are dominated by room modes at low frequencies which needs to be addressed either with passive room treatments or a multi-subwoofer approach. It has been shown that early reflections can be beneficial for perceiving ambiance and spatial cues in the music.

This article and accompanying YouTube video discusses these topics in more depth and gives recommendations for consideration of sound reproduction in small room acoustics.

Read: Early Reflections and Bass for Small Room Acoustics

 

[gene]

 

[shadyJ]

The paragraph about the usefulness of multiple subs in small rooms needs to be bolded, capitalized, and then embossed into a sledgehammer which is then used to beat over the head of everyone who decided one subwoofer was enough when they designed their recording studio, mixing room, or any kind of critical listening room. Audiophile subjectivists are especially bad with this, as they are with everything else.

I do have one question about a statement: "Since room modes are minimum phase phenomena, the amplitude response contains ALL of the essential information. That is not a matter of opinion, it is physics. If a curve is made more smooth by any means, the resonance is attenuated in both frequency and time domains." We are having a discussion about group delay and decay times in bass frequencies in this thread, and one thing I am not one hundred percent sure about is group delay always going to show up in a frequency response chart, or can you have a flat response but still get uneven decay between frequencies? The statement, "If a curve is made more smooth by any means, the resonance is attenuated in both frequency and time domains" seems to suggest that a flat response would also mean even decay times.

Anyway, great article, lots of good data, and the idea that a single sub is all that is needed has to be jettisoned by anyone who is serious about getting good sound.

 

[gene]

The paragraph about the usefulness of multiple subs in small rooms needs to be bolded, capitalized, and then embossed into a sledgehammer which is then used to beat over the head of everyone who decided one subwoofer was enough when they designed their recording studio, mixing room, or any kind of critical listening room. Audiophile subjectivists are especially bad with this, as they are with everything else.

I do have one question about a statement: "Since room modes are minimum phase phenomena, the amplitude response contains ALL of the essential information. That is not a matter of opinion, it is physics. If a curve is made more smooth by any means, the resonance is attenuated in both frequency and time domains." We are having a discussion about group delay and decay times in bass frequencies in this thread, and one thing I am not one hundred percent sure about is group delay always going to show up in a frequency response chart, or can you have a flat response but still get uneven decay between frequencies? The statement, "If a curve is made more smooth by any means, the resonance is attenuated in both frequency and time domains" seems to suggest that a flat response would also mean even decay times.

Anyway, great article, lots of good data, and the idea that a single sub is all that is needed has to be jettisoned by anyone who is serious about getting good sound.

Ideally yes but it's always better to eliminate loudspeaker resonances (even at low F) anechoically so you can separate the room influence from what's actually happening in the speaker.

 

[Floyd Toole]

The group delay/phase shift in woofers and subwoofers is determined by the shape of the high-pass characteristic of the loudspeaker itself. These are minimum-phase systems: the transducer with its enclosure. The lower the slope of the high-pass characteristic, the less dramatic the accumulation of phase shift. This can be changed with a parametric equalizer, making it better or worse - based on anechoic data, not room curves. Some people interpret this as a condemnation of bass reflex systems, but that is only true if the system cuts off in the middle of the musically-interesting frequency range. Tuning a reflex system to below 25 Hz or so, moves the problem well below the kick drum "punch" (around 60-80- Hz) or "snap" even higher frequencies. One example I use is Paul Simon's "Graceland" - great music, and Steve Gadd's kick drum is as "tight" as you would want. Why? There is very little in the recording below about 100 Hz. Result: it sounds similar on almost everything. But there is no real bass. Take away those pesky fundamental frequencies and what is left is "tightness". But a real kick drum has bass.

Another example, the Sheffield "Drum Record" which I used to use as a test - it was a direct-to-disc recorded LP, remember those? It was great, but a very good cartridge was needed to track it without gross distortion. Then came the CD version. Suddenly it was not as impressive. What's going on? Well, it turns out that they used an analog tape recording - a backup - running at 30 ips as the master for the CD. Analog tape at 30 ips has very little output below 50 Hz; it is a sharp high-pass (phase shifting!) filter. An A vs. B comparison between the LP and the CD was a convincing win for the LP. LPs are absolutely inferior to CDs when the masters are identical. In this case, the "masters" were not identical.

There is a ton of group delay/phase shift in recordings created by high-pass filters in the mix to get rid of HVAC rumbles, to prevent woofer overload (yes in movies too), microphones that do not go down to DC, analog tape, and on and on. A perfectly phase correct sub or woofer will not always sound 'perfect', whatever that is.

And we haven't even started on the contribution of the room.

 

[shadyJ]

One example I use is Paul Simon's "Graceland" - great music, and Steve Gadd's kick drum is as "tight" as you would want. Why? There is very little in the recording below about 100 Hz. Result: it sounds similar on almost everything. But there is no real bass. Take away those pesky fundamental frequencies and what is left is "tightness". But a real kick drum has bass.

Is the impression of bass in that recording is due to the phenomenon of the 'missing fundamental'?

Another thing, I have a thought that many times when people say a subwoofer has 'tight' bass, what they are really talking about is mid bass, and that often when a sub has formidable deep bass output, it is thought to sound 'slow'. I think this perception can work against systems with a flat frequency response in bass, and if one were to elevate the mid bass frequencies, say 60 or 70 Hz+, the system's bass will be thought of as 'quick' or 'tight' or 'musical'. I am not saying it is a mistake to think this because the matter is entirely subjective, but, of course, it would be a mistake to call such a system accurate. Do you think there is truth to this or is my thinking off here?

 

[Floyd Toole]

The missing fundamental is almost certainly a factor from the perspective of pitch, but not "kick in the gut" bass energy. It is not there.

Loudspeakers with substantial low-frequency output energize room modes, which make things sound "slow". Tame the modes, and things speed up considerably. For many years the Brits had a love affair with small bookshelf loudspeakers, on stands. They were even the "reference" loudspeakers for some prominent journalists. Why? No low bass = no boom in the often thick plastered or masonry walls in many homes. Rather than address the problem, they seemed to accept a compromise - insufficient low bass.

But, there is more. As I discuss in great detail in my new 30 page AES paper - a free download - bass is not constant among recordings. Figure 2.4 in my book is a scary look inside recording control rooms. So I really believe that tone controls are essential if you are fussy.

An anechoic flat on-axis speaker will not be flat in a room - the bass will rise. The task of multi-subs, bass traps, and equalization is to get rid of resonances. Once that is done you need to find a suitable bass level. It very likely won't be flat, and it cannot be the same for all recordings or movies. Hence the tone controls. The problem is that the industry has no standards that work, and most recording engineers think they know better . . .

in my room the bass is very smooth and flat down to about 23 Hz, still going at 17 Hz. Figure 13.18 in my book. Five listeners hear almost exactly the same bass and it is clean, deep and 'tight'. You can walk around the room and the bass is remarkably consistent. It should be ideal. Yet, when I listen to music in the foreground, I sometimes have to resort to tone controls. The system is fine, the recordings are what they are - not all the same, and not all good.

 

[jim1961]

I think the goal is the same regardless of the size of the room. A smooth direct response coupled with a smooth room response. By room response, I mean the measured energy after T=0. This can be seen in decay, spectrograms and waterfall graphs. That is to say, at whatever time interval your looking at, the frequency response should be smooth and relatively equal.

The distinction between small rooms (generally those <5000 cu ft) and large rooms gets lost given 98+% of those frequenting forums like this have small rooms.

What might be interesting, is whether 3500, 2500 and 1500 cu ft rooms (all considered small) should share the same approach in regards to treatment.

 

[Floyd Toole]

Yes, the goal is the same - get rid of resonances: in the room at low frequencies, and in the loudspeaker at all frequencies. When that is done, both the frequency domain info and the time domain info are agreeably well behaved. Beyond that, do what is possible by choosing good loudspeakers and not messing them up with inappropriate "acoustical treatment" so that reflected sounds resemble the direct sounds. In small rooms the direct sound is not the dominant factor in what we measure or hear, except at quite high frequencies. My new AES paper explains this in detail.

All of the rooms you mention will have resonance problems, and taming them will involve the same alternative methods and decisions.

Most waterfalls I see lack sufficient time resolution to reveal what is actually happening in that domain - see Figure 13.23 in my book. There is a trade-off between frequency resolution and time resolution - you cannot have both. A lot of published data obscures events in the time domain, showing what looks like ringing when the reality can be very different.

 

[jim1961]

What I meant by different treatment approaches for different sized rooms was the actual implementation. For instance a 1500 cu ft room may limit the amount of diffusion you could utilize and require a higher percentage of absorbent compared to a larger small room (3500 cu ft) where a larger amount of diffusion may be possible and somewhat less absorbent.

As far as Time vs Freq. resolution, could one split things into two views. One optimized for Time, the other Frequency. Then, looking at both, derive a clearer view?

 

[TheWarrior]

The waterfall Floyd is citing also includes Relative Level dB as a third axis, so changing the perspective of a three dimensional graph would have you looking at the time domain like incoming ocean waves. How deep is the trough between each crest?

On the previous page, "However, because we know that low-frequency room resonances generally behave in a minimum-phase manner, we know that if there are no prominent peaks protruding above the average spectrum level, there will not be prominent ringing in the time domain. It is this indirect, inferential knowledge that permits us to confidently use frequency responses as a primary source of information about room behavior at low frequencies."

 

[Floyd Toole]

Thanks Warrior, I forgot I said that. Still true, too

Those who insist on waterfalls to prove a point don't understand the physics of the situation. I think I have been around this course too many times to do it any more. It is all in the book. It is not a personal opinion.

Diffusers make absorbers work harder, making them more effective, and requiring less area. However, recent papers are showing that a reflection "diffused" by a QRD (an engineered surface) are not less audible than that from a flat reflecting surface - it seems that we perceptually "integrate" the time-distributed energy. Interesting. So much to learn, so little time . . .

 

[Floyd Toole]

jim1961 Sorry, I just realized that I didn't answer your question. If you do a steady-state (very long time window) frequency response you will see the frequency-domain information in maximum detail, but nothing in the time domain. This will show evidence of audible resonances, and, as I said, by inference what the time domain will look like. Then shorten the time window, thereby reducing the resolution in the frequency domain - the curve at the back of the waterfall will be much smoother, but now some detail will begin to show up in the time domain. At some point you will begin to see big changes in the time domain, but with low resolution in the frequency domain. Figure 13.23 in the book illustrates this. In the example there are no problem resonances, but some of the maladjusted waterfalls suggest that there are. So, yes you can find ways to get a clearer view in the time domain, but if you do it as many times as I have, you eventually begin to trust physics - and your ears

 

[TheWarrior]

However, recent papers are showing that a reflection "diffused" by a QRD (an engineered surface) are not less audible than that from a flat reflecting surface - it seems that we perceptually "integrate" the time-distributed energy. Interesting. So much to learn, so little time . . .

Interesting indeed. Does our 'perception of a diffused sound wave's integration' include localization?

 

[BoredSysAdmin]

If I may be so bold and quote from one of my favorite books - Sound Reproduction Loudspeakers and Rooms, just to help on clean on concept of what is a "small room"

Diffuse-field theory may not apply perfectly to concert halls, but it applies even
less well to other kinds of rooms. In the acoustical transition from a large performance
space to a “small” room, it seems that the significant factors are a
reduced ceiling height (relative to length and width), significant areas of absorption
on one or more of the boundary surfaces, and proportionally large absorbing
and scattering objects distributed throughout the fl oor area.

Very rough translation is - if you don't place your speakers in concert hall or at least a huge warehouse, but instead in one of rooms if of your own home - most likely that room is "small" from acoustics factors standpoint

Please correct me if I wrong. This implies that the information in the article applies to 45x30x10 rooms just like it does in 15x10x8 (all in ft)

 

[TheWarrior]

Absolutely! I don't have a calculator handy, but even that 'large' of a room is only about 13,500 cu ft. You're not worried about balanced sound across 100 seats in a room that size. Maybe 3-4 rows at most!

 

[Floyd Toole]

Reverberant concert halls and auditoriums are acoustically "large" in that they have a significantly relatively diffuse sound field because the room boundaries above the audience are reflecting and diffusing, and the volume is made large - high ceilings - to encourage reverberation. Reverberation times are 1.5 to 2 s or more. These are performance spaces for musical sources that do not have volume controls - the precious energy must be preserved and distributed to the audience.

Cinemas, including large 500 to 1000 seat venues also have audiences covering the floor, but the side walls and sometimes portions of the ceiling are covered with absorbing material, often 2-inch fiberglass or the like. These are acoustically "small", with reverberation times not very different from our home theaters: 0.3 to 0.6 s. However, unlike an "omnidirectional" orchestra, we have moderately directional loudspeakers. Above about 200 Hz all significant acoustical events happen in the first 50 ms - direct sound plus a few early reflections. Above about 800 Hz the audience is in the direct sound field. My recent AES paper explains this in detail. The sound field is absolutely not diffuse, and we don't want it to be.

The only other factor distinguishing geometrically small rooms is the presence of standing waves at low frequencies. These are substantially predictable in rectangular rooms, which is why such shapes are desirable.

 

[BoredSysAdmin]

Reverberant concert halls and auditoriums are acoustically "large" in that they have a significantly relatively diffuse sound field because the room boundaries above the audience are reflecting and diffusing, and the volume is made large - high ceilings - to encourage reverberation. Reverberation times are 1.5 to 2 s or more. These are performance spaces for musical sources that do not have volume controls - the precious energy must be preserved and distributed to the audience.

Cinemas, including large 500 to 1000 seat venues also have audiences covering the floor, but the side walls and sometimes portions of the ceiling are covered with absorbing material, often 2-inch fiberglass or the like. These are acoustically "small", with reverberation times not very different from our home theaters: 0.3 to 0.6 s. However, unlike an "omnidirectional" orchestra, we have moderately directional loudspeakers. Above about 200 Hz all significant acoustical events happen in the first 50 ms - direct sound plus a few early reflections. Above about 800 Hz the audience is in the direct sound field. My recent AES paper explains this in detail. The sound field is absolutely not diffuse, and we don't want it to be.

The only other factor distinguishing geometrically small rooms is the presence of standing waves at low frequencies. These are substantially predictable in rectangular rooms, which is why such shapes are desirable.

Thank you Dr. Toole for detailed explanation.

All I wanted to highlight that "small" room in context of our site typical reader - is ANY room in their house or apartment.

 

[jim1961]

Reverberant concert halls and auditoriums are acoustically "large" in that they have a significantly relatively diffuse sound field because the room boundaries above the audience are reflecting and diffusing, and the volume is made large - high ceilings - to encourage reverberation. Reverberation times are 1.5 to 2 s or more. These are performance spaces for musical sources that do not have volume controls - the precious energy must be preserved and distributed to the audience.

Cinemas, including large 500 to 1000 seat venues also have audiences covering the floor, but the side walls and sometimes portions of the ceiling are covered with absorbing material, often 2-inch fiberglass or the like. These are acoustically "small", with reverberation times not very different from our home theaters: 0.3 to 0.6 s. However, unlike an "omnidirectional" orchestra, we have moderately directional loudspeakers. Above about 200 Hz all significant acoustical events happen in the first 50 ms - direct sound plus a few early reflections. Above about 800 Hz the audience is in the direct sound field. My recent AES paper explains this in detail. The sound field is absolutely not diffuse, and we don't want it to be.

The only other factor distinguishing geometrically small rooms is the presence of standing waves at low frequencies. These are substantially predictable in rectangular rooms, which is why such shapes are desirable.

Talking about reflections, I am curious what bandwidth is necessary. That is, referring to the pic from your book, am I interpreting this correctly to mean that reflections in the 1K-8K range are the most responsible for our sense of room spaciousness or liveliness?

Are reflections below and above this frequency range still important if one is trying to bring more life to a room that is on the dead side (rt60 = .16 or so) ?

What i am trying to reconcile, is the idea that reflection content should be smooth and frequency balanced which implies reflections content should extend down to the transition frequency ideally. Yet, my first point and the attached graphs, suggests the transition to 1K area is less important perceptually.

 

[Floyd Toole]

Catching up with some responses . . .

Warrior: I don't know about localization precision with QRD reflections. More research, I guess. Because these are secondary to the direct sound, normally suppressed by the precedence effect, it might be a moot point. If the precedence effect breaks down for some reason, the the reflections will be localizable, but the most probable reason for breakdown in our small rooms is that the spectrum of the reflection is different from that of the direct sound. - i.e. either bad off axis loudspeakers being specularly reflected, or reflections that are spectrally distorted. That is why I tend to favor acoustical treatments of two kinds: absorb it all, or reflect it all. Anything else adds risk of degrading good loudspeakers. That said, there are some manufacturers of engineered diffusers who publish reflection spectra for their devices. If that is good, they are an option, remembering that we now know that they may not change the audibility of the reflection as we once thought. Confusing, eh?

Boredsysadmin: yes, any room that we are likely to have in our homes is small, both acoustically and geometrically.

jim1961: Wow, 0.16 s. That is truly dead - too dead for comfortable conversation for most people. Why?

Anyway, if that is a fact of life and you want to add some sense of space while the playback is occurring, some reflecting surfaces will be needed, placed to pick up the early reflections (contrary to much folklore and professional audio practice). The "space" information you are after cannot be the small room you are in, it is in the recordings (or not, as the case may be). Local reflections won't do much to add "space" because the reflections occur too early (reflections generating interesting spatial effects in large rooms occur 60- 100 ms after the direct sound - you will do well to get 1/10 of that.)

My suggestion - hold your nose - try a multichannel upmixer. There are several versions, and they vary considerably. The old Lexicon Logic 7 was one I used to inject some spatial life into certain recordings. It mostly left the soundstage alone and added some spatial envelopment that was nice. It was totally program dependent, because no recording engineer uses them to check their mixes. When it worked, switching back to "spatially deprived" stereo was a real comedown. I no longer have it because my processor became obsolete, and now I make do with the normal options, which change with my mood. It is entertainment, not replicating a concert hall. For that we need multichannel, which, sadly, the music folks don't use. Good luck.