Bruce Black of MediaRooms Technology LLC
Did you know that your real-time analyzer is lying to you? Yes, you. You've been loyal to it, you've treated it kindly, feeding it only the best power, rack-mounted it with the finest screws, and it has betrayed your confidence and loyalty. How typical.
A real-time analyzer is, of course, intended to give you a moment-by-moment picture of the audio frequencies in the room. That data can be used by the engineer for tuning the sound system in a recording studio or live performance, compensating for room resonances or frequency dead zones. The assumption that the analyzer is giving you everything you need to know to accomplish this is false, according to Bruce Black of MediaRooms Technology, LLC.
On Tuesday, April 24, Bruce presented his case for paying attention to the dimension that real-time analyzers (RTAs) don't show: time. In a fascinating presentation, Bruce explained how having multiple frequency spectrum samples, taken at the same location over time, can reveal characteristics of the room that may be difficult to capture with an RTA.
Once upon a time, tools like the Klark Teknik DN60 spectrum analyzer were the best available to show what sound was doing in a particular room. Nowadays, software like Audio Tools for the iPad, have become the standard for their resolution and convenience.
We tend to make assumptions that the information we glean from these devices is complete and comprehensive enough that they give all the information needed to tune the room flat, thus producing "accurate, ear-pleasing performance." "But," he said, "there are characteristics that none of these RTAs reveal. Some of these characteristics may have an even greater influence on the subjective listening experience than the 1/3 octave frequency response the RTA gives us, and may be dynamic and change over time." It could also be that the graphic EQ is not equipped to fix the issue.
Mr. Black showed frequency analysis plots of two spaces he had worked in. The graphs extended from about 10 Hz to 24 KHz without any smoothing. The first was a mastering studio. The graph had a reasonably flat response, with a pair of peaks at 100 Hz and 300 Hz (these are the room resonances), and a sharp dip between them at 110 Hz. This room was "treated" by applying "the so-called common wisdom", not necessarily scientific analysis.
The assumption is that the flatter the room response, the more accurate the room will reproduce a recording during playback. Thus, the overall goal is to flatten the room's response. Smoothing the curve to a 1/3 octave response, what human ears are supposed to equate to, the curve looked even better, and still showed the resonance peaks. However, when a recording was played back in the room, it still sounded bad.
For smaller spaces, the lower frequencies provide the bulk of the sonic quality because of the small number of resonances. When zoomed in on the 10 Hz to 1000 Hz of the unsmoothed wave, resonances could be seen at 32, 60, and 85 Hz. Calculating the wavelengths of each produces lengths that correspond nicely with the physical dimensions of the room. Still present was the dip at 110 Hz, which looked a lot like phase cancellation. When the microphone was moved about the room, the dip nearly disappeared. The room also had a gentle roll-off below 80 Hz, probably due to the playback speaker's capabilities. All is well, right?
Well, not quite. The human brain "integrates the continuous sonic input" instead of just using a snapshot. "Time is a very important thing that very frequently does not even occur to us when we evaluate a room's acoustical performance, yet it is critical to a room's sound, perhaps even more than frequency response."
When the mastering room was evaluated at different instances in time, several items become noticeable. At 120 milliseconds after the initial sound, the room's physical characteristics started to exert influence: the higher frequencies decayed around 10 dB, the resonances remained strong, and a -25 dB dip at 46 Hz has appeared. This dip was lower than the natural test speaker rolloff, so it was from the room's response.
At 240 milliseconds after the direct sound, the 110 Hz dip is now -40 dB from its initial level, highlighting the resonances on either side of it. This is only one-quarter of a second after the pulse. At 360 milliseconds, the frequencies on either side of the 100 Hz resonance were about 50 dB down from their initial levels. Imagine what this sort of response has on bass guitars, kick drums, or tubas. And EQ is incapable of fixing this, since the response is dynamic with respect to time; one would need to start with a 50 dB bump at the initial pulse. It also appeared that the room resonances (110 Hz and 300 Hz) were poking up through a larger dip, suggesting the room was to blame by absorbing low frequency energy. At 480 milliseconds, the levels across the lower spectrum have started to recover, but only by about 7 dB. Remember, all this happened in less than a half second. The overall sound has decayed around 45 dB.
By applying well-known principles of physics and acoustics, then plotting room response graphs with frequency, amplitude, and time data, Mr. Black demonstrated the discrepancy between one's ears and the real-time analyzers' displays.
Mr. Black went on to describe several other rooms with different problems: a significant bump at 200 Hz remaining past a half second, or an overdamped room that loses 80 dB above 68 Hz in less than a quarter second (classic boominess).
Some of the problems involve the materials and construction techniques used to create the rooms. Using the more relaxed residential standards instead of the more stringent acoustical standards can create un-intentional absorption or reflective panels and wall spaces.
"To one degree or another, the frequency and intensity that the panel vibrates is determined by the real-world mass, density, and stiffness of the vibrating material, the density and stiffness of the framing it is screwed into, how the laborers installed each panel, how many screws the laborers used for each panel, how much power was used to drive the screw determining the force which holds the panel, the physical qualities of the screw: hardness, thread depth, etc, and the real-world distance between the framing members. With all the possible variations in materials, construction, and installation technique, you can see each vibrating panel is a unique animal, and vibrates at its own individual, unpredictable frequency and intensity."
One of the best solutions, he said, was the installation of more drywall screws, adding to the stiffness of the wall's drywall panels. They are inexpensive, don't require batteries, and can be placed no farther than four inches apart. Additional solutions include increasing the panel thickness, using drywall adhesive both between the studs and the drywall, and between the layers of drywall.
We want to thank Mr. Bruce Black for sharing his insights on well-designed acoustical spaces and the problems that don't show up with a simple real-time analyzer. Paying attention to time can reveal the true symptoms, which in turn can lead to the correct solution, which may be stiffening the walls with more screws.
