My Music Room
photograph of the interior (105 kb image).
The topics in this section are soundproofing, and wall treatments to reduce reflections.
A major objective of my music room project was to make the room soundproof. The primary motivation for this is so I can blast my music at whatever volume level my little heart desires without worrying about putting my marriage in jeopardy. Two other benefits are: (1) I don't have to worry about disturbing the neighbors, and, (2) external noises like leaf-blowers, etc., don't disturb my listening experience. The actual end result: when my music is real loud my wife can hear it one floor below, but two floors below in her office it is virtually inaudible. Our marriage is strong enough to survive under these conditions.
Acknowledgment: almost everything I know about soundproofing comes from the Handbook for Sound Engineers, Chapter 4 (see references). My son found another web site that contains good information. Various soundproofing products are marketed by Quiet Solution.
A fairly common misconception is that soundproofing is only achieved by brute force - i.e. concrete walls. In fact, an air gap is the best way to achieve soundproofing. The Handbook states that 3 dB more sound gets through an 8-inch concrete block wall than through the gypsum and resilient channel wall construction discussed below. A solid concrete wall 12-inches thick is about 10 dB better than hollow 8-inch blocks, but there are easier and cheaper ways of getting equal performance.
Ideal soundproofing requires total mechanical isolation of the room from the rest of the house. In other words, no structural connections. This is rarely possible. The next best thing is connecting only through isolation mounts such as Neoprene "hockey pucks." In my situation this was also impractical.
The room was built using two main soundproofing techniques: (1) staggered studs, and (2) resilient channel. Staggered studs means that there is one set of studs supporting the inside wall, and a separate set supporting the outside wall. Thus when the inside wall vibrates, and vibrates its set of studs, the 2nd set of studs and outside wall are relatively unaffected. The staggered pattern can be seen in the photo (33.2 kb) at an early stage of the construction. This technique was used on all of the new walls. Thicker than normal Fiberglass insulation was woven between the studs. For both the ceiling and interior walls, resilient channel was also used. This means a layer of gypsum attached to the studs, then the resilient channel - flexible metal strips - and finally another layer of gypsum. The resilient channel is shown in the second construction photo (16.4 kb) prior to adding the 2nd layer of gypsum. Here the idea is that the flexibility of the metal strips reduces the vibration coupled from the outside layer of gypsum to the inside layer of gypsum. According to the Handbook the staggered studs plus resilient channel should cut the sound transmission by more than 50 dB. If I am playing music at a loud 100 dB SPL, outside the wall it will be 50 dB SPL. This is about 10 dB above the background noise level in a quiet room, and would be audible, but muted.
I tested the construction by reversing the normal situation: playing loud music in the main part of the house, and using a wine glass as a stethoscope to see where it was coming into my music room. The staggered stud plus resilient channel was terrific. There was one pre-existing wall in common with the main part of the house where only resilient channel could be used. The sound transmission was by far the highest for this wall, but still OK. The real weak link at this point was the bare floor, which was the roof of the pre-existing garage, where sound just poured through.
Other construction details are that double-glazed windows were used, and a heavy solid door with tight weather-stripping. The room has a separate heater to avoid connecting to the main house heater ducting. The ventilation system uses flexible ducting rather than sheet metal. The 3-foot long air inlet duct has two bends; the outlet duct is straight, but 6-feet long. All of these things worked great. I also caulked around the edge of the floor and filled electrical outlets, etc., with polyurethane foam. I am not convinced either of these things were worth the effort, but it is difficult to know for sure.
Covering the floor with two layers of 30# roofing felt (to dampen vibrations), a 1/2-inch layer of rebond carpet pad, and heavy carpeting improved things dramatically. But the floor is still the weak link, and in the garage below the music can be heard quite clearly. Fortunately the wall between the garage and the house cuts it down a lot.
For folks that do not have the luxury of building a new room, the resilient channel can be used to retrofit an existing room. This will definitely make an improvement, but my experience says that you shouldn't expect too much. If soundproofing is really important, about the only way to obtain a successful retrofit is build a new shell inside an existing room, with construction techniques similar to those described here.
Back to the top
Room acoustics can be adjusted by adding soft furniture and heavy drapes, or by the use of dedicated sound absorbing materials. Diffusers can also be used, which break up the reflections from walls and scatter the sound over a wide angle. All of these techniques are thoroughly described in The Master Handbook of Acoustics (see references). In this section the use of absorbing panels to eliminate unwanted reflections is described. Another type of absorber is a "tube trap" which does roughly the same thing as a lot of soft furniture. Another music room on the web makes heavy use of tube traps. The wall treatments described here are designed to improve the acoustics at the higher frequency range of the audio band. Low frequency room acoustics are discussed in a separate section.
As discussed in the section on how we locate the source of sounds, reflections that reach the ear within 0.1 to 1.0 milliseconds of the direct sound arrival have the most serious effect on the apparent location of the sound source. The effects are also most serious for reflections in the horizontal plane. Since source location is the basis of stereo imaging, the ideal situation is that only the sound coming directly from the loudspeaker reaches your ears in this critical time interval. There are also dozens of later reflections that reach your ears with delays between 1 and 300 milliseconds - see the section on room acoustics. These later reflections give the room its "dead" or "live" character, and have little or no effect on stereo imaging.
I am going to go into a fair amount of detail about reflections. The reason for this is to show how a little detective work can identify individual reflections and quantify the improvement obtained by wall treatments. The technique involves some fairly simple geometrical calculations combined with the measured system impulse response.
An illustration of an early sound reflection is shown in the music room floor plan (4.2 kb). This is a view looking down at the room, the rectangular outline being the four room walls. The speakers are on the left, the driver locations being indicated by the small + symbols. The circle near the middle of the room is the "sweet spot," where the sound quality is optimal. The blue line shows the direct sound path from the left speaker to the listening position. The solid red lines indicate the paths of reflected sound, which appear to emanate from an "image" loudspeaker outside the room boundary, as indicated by the dashed red lines. There are many such images created by wall reflections as described in the section on image analysis. Here we are dealing with just one, which produces the earliest reflection. The total length of the solid red paths from the speaker to the line ends are all equal, and 23.5 inches longer than the direct path. This corresponds to a delay of 1.74 milliseconds relative to the direct path, so it is actually outside the critical time interval of 0.1 to 1.0 millisecond. The really problematic reflections, within the critical time interval, have all been eliminated by placing the speakers far enough away from the walls.
Diffraction problems have also been minimized by the design geometry. The equipment shelves between the two loudspeakers are flush-covered with glass doors (air circulates behind the equipment), and the other sides of the speakers are flush with the wall. Thus, the problematic diffraction from speaker cabinet edges has been eliminated by the simple expedient of eliminating the edges (in the horizontal plane). Edge diffraction in the vertical plane is less serious since humans have a poorer sense of location in the vertical plane. The vertical edges are 2-1/2 feet away from the nearest driver anyway. Diffraction from the front panel of the speakers is a topic in itself, and is discussed in the section on speaker construction.
Even though the earliest reflection is outside the critical time region, I decided to reduce it. Again, geometry is the first line of defense. By angling the speaker faces inward, the speakers point directly at the "sweet spot," and the angle of the reflected path leaving the speaker is increased with respect to the direct path. The sound from all cone speakers is strongest in the direction where it points, and becomes weaker as the angle is increased away from this direction. Thus, the reflected sound is already reduced quite a bit by angling the speaker cabinet. Also see Lynn Olsen's comments on toe-in. The second step is to place a 24-inch wide sound-absorbing panel in the location indicated by the short yellow line along the wall in the floor plan. This is the spot on the wall where the sound is reflected towards the "sweet spot." The panel can also be seen in the photograph of the interior (same 105 kb image as above). I used 3-inch thick absorber panels from Auralex, which I am quite happy with. A second panel was located in a similar position with respect to the right speaker.
The other early reflections come from the floor and ceiling. The floor reflection is considerably reduced by the carpeting. Two absorbing panels were placed on the ceiling to cover the area that reflects sound towards the "sweet spot."
I made two impulse response measurements with my CLIO sound measurement system, one with the absorbing panel up, and one down. For a better comparison, I exported the resulting measurements into Matlab, and produced a plot with the two measurements overlaid. (42.4 kb). The horizontal axis is time in milliseconds. The vertical axis is the sound pressure measured by the calibrated microphone, located at the "sweet spot" position of the floor plan. The time origin has been shifted such that the peak of the spike of the direct sound from the speaker is at t=0. The impulse from the direct sound reaches a peak value of 3.39. (The units of this scale are arbitrary - we only care about relative values). The red curve represents the response with the absorbing panel removed, and the black with the panel present. The vertical blue line is at 1.9 milliseconds. The red spike below this line is the reflection from the wall without the absorber panel in place. The reflection is 0.16 milliseconds later than expected, indicating a path that is 2 inches longer than the 23.5 inches calculated using the floor plan. (Engineers are used to encountering small discrepancies like this, but I am always happier when it comes in on-the-money). It is clear that this is the wall reflection, because the black curve with the absorber present reduces this spike from a level of 0.859 to 0.365 (is this cool stuff, or what?). The area around this spike is the only place the red and black curves are significantly different, as would be expected. Converting the numbers to a dB scale, the wall reflection is down 19.4 and 11.9 dB relative to the direct sound, with and without the absorber panel in place, respectively. According to a curve on page 61 in The Master Handbook of Acoustics (see references) a reflection 19.4 dB down and delayed 1.9 milliseconds is inaudible. At 11.9 dB it is well within the audible range.
Another geometric calculation shows that the small blob around 1.5 milliseconds is the reflection from the floor. The carpet absorbs high frequencies better than low frequencies. The absence of the high frequencies turns a spike into a blob. The spike just following the blob is a mystery. I haven't a clue where this reflection is coming from. (Mystery solved! See footnote). Measured and computed data on reflections arriving later in time are presented in the section on room acoustics.
The ceiling reflection would show up at 2.7 milliseconds if the absorbing panels were not in place. The ceiling reflection has also been significantly reduced, as can be seen by making a comparison with the data in the section on room acoustics. The measurement here and in the room acoustics section used a different microphone placement, so the time of arrival of the ceiling reflection differs, but the relative magnitudes can still be compared.
So the bottom line is that the dominant wall reflection was reduced by 7.5 dB, making it inaudible, by putting up an absorbing panel. The Auralex catalog indicates that their 3-inch absorber absorbs essentially 100% of the incident sound above 500 Hz, 50% at 250 Hz, and 23% at 125 Hz. The 7.5 dB reduction corresponds to a power absorption of 82%. My guess is that their measurement is for sound that is aimed directly at the panel; the reflection we have been dealing with glances off the panel at an angle. I am quite happy with 82%.
My wife Cynthia tends to ask me awkward questions like: "But can you hear any difference?" In this case, damned if I know - I didn't even try an AB listening test. Sometimes I just try to do anything I can to make a design better, and the end result is usually good.
Footnote: Thanks to James Galloway for solving the spike mystery (August 2006). I worked hard to time-align my speakers, so the sound from woofer, midrange, and tweeter arrives simultaneously at the sweet spot where the mike sits. This is the reason that the large spike at t=0 is quite sharp. The alignment only works for a point at a fixed height relative to the speaker. The image speaker in the wall appears to be at the same height as the real speaker, the image sound is also time-aligned, and the spike at 1.9 ms is also sharp. What didn't occur to me when I originally wrote this is that the image speaker in the floor appears to be under the floor, and it is not time-aligned with the mike. The blob is a floor reflection, but only from the midrange. It is spread out not (or not only) because the carpet absorbed the highs, but because the crossover removes the highs. It was a first-order crossover at the time of this measurement, so the spreading is modest. The spike is a floor reflection from the tweeter, and since it contains all of the high frequencies it is quite sharp.
Back to the top
To the table of contents.