Equalization is the easiest and the hardest part of my job tuning sound systems. Locating peaks and dips, and bringing them into submission is child’s play compared to making sure the speakers are aimed, splayed, spaced, delayed, and level set for even coverage. The hard part about equalization is the guessing game. Do they want it flat? If not flat, then what? Cinema has an X Curve. Everything else in the industry uses the “read my mind” curve. I recently attended a presentation where I learned the answer to the question I have been grappling with for 30 years: a 12dB slope from lows to highs works in all occasions. My mind flashed back through a thousand tunings and concluded indeed that +12dB is indeed the average between my clients that like it flat and the ones that go for +24dB in the low end. We are within ±12dB of a one-size-fits-all answer.
Figure 1: The transmission path of sound from source to listener. Notice that nearly all mechanisms lead to frequency response tilting over distance. Click here to see a larger image.
Where does this all come from? Why does “flat” wear a white hat on your console, amplifier, cables, and even individual speakers, and a black hat as soon as these components are fully assembled? How comfortable are you with me inserting a cable with a 12dB LF to HF slope on a flat system if I can get you the same frequency response as a flat cable on a tilted speaker system?
Where does the urge to tilt come from and why do we save it for last? It’s a product of the natural world and how we experience sound sources over distance. The tilt in a frequency response is a primary factor in how we determine how far away we are from a source. The guessing game of how much tilt to put in a system is the artistic decision of how close we want the audience to feel to the sound system (and the signals we are sending through it).
There’s “Equal” in EQ
It is obvious that the term “equalization” implies that we are compensating for the inequality in the speaker response with inequality in the electronic filter response. Sorry mom: two wrongs do make a right. A peak in the speaker gets a dip filter of the same center frequency, bandwidth, and dB level. The result is a peak flattened down to the nominal level. The combination of speaker/room and filters is flat, i.e. equalized. Saying that you equalized it flat is redundant. If it’s not flat, it’s not equal, and if it’s not equal, it is certainly not equalized.
We know in practice that our industry is fast and loose with terminology, and this is no exception. It is rather amusing to note that equalizing something to equality (i.e. flat) is considered an aberration. Everybody knows the last thing we want to do is put too much equal in our equalization. I guess it’s like “fish is too fishy.”
Figure 2: A loudspeaker measured indoors at progressive doubling of distance. Note the increasing spectral tilt, which clues the listener to the speaker’s range. Click here to see a larger image.
Fear of flatting is a common phobia in speaker treatment but not so in other parts of the system. For example, if you saw the HF rolled off 1dB in the electronic signal path, would you leave it that way or equalize it (i.e. make it flat)? It’s difficult to argue against flat in this case but this goes out the window once sound gets into the air.
Sonic Image Range
Our perception of sonic image is multifaceted. We immediately think of vertical and horizontal localization but there is a third dimension—depth. That’s where equalization fits in. The frequency response of any sound source transmitting through the atmosphere of Earth changes over distance. The response tilts in favor of the lows over the highs. The reasons are A.) the atmosphere (loss of HF response in the air) and B.) Earth (reflections off the ground and other surfaces favoring the LF response). A “normal” violin sound in the near field is very different than far away. If it only got quieter over distance, we could easily fool our sonic range finder, but there is more to it. Mezzo forte sounds louder, but not closer than pianissimo. The normal near response has more HF (less air) and less lows (less room) compared to a level-normalized midrange. I have always found it amusing that when people ask for “more air” in their mix, they are looking for more VHF extension, which is exactly what adding real air to the transmission takes away.
Perceived sonic range is a combination of spectral tilt and direct/reverberant ratio clues. Spectral tilt in favor of the lows and/or rising reverberation moves the perceived depth away. The on-axis response of a distant source (compared to its near field response) will be unequalized in favor of the lows (perceived as far away). The near field off-axis response of the same source will also be unequalized (due to being off axis). It is perceived as far away even when close (due to the HF loss). A natural sound source can be perceived at its actual depth or further away (off axis or strongly reverberant).
Loudspeakers have the unique capability of sonic range ventriloquism. Unlike the violin, they can be perceived as closer than they actually are. A sound system that’s “in your face” is one that is perceived as closer sonically than visually. This is accomplished by equalization and reverberation control. A flatter and more directional speaker is more likely to create an up close and personal experience, while heavy tilting and wide directionality is most likely to sound distant.
Linking Equalization and Sonic Image Depth
We don’t have an image depth perspective for a length of wire. That’s why there are no objections to equalizing any losses it accrues and returning its response to flat. Speakers always have an image perspective, as do the input signals they are transmitting. Let’s compare two paths to the listener: Direct sound from a natural instrument (e.g. violin) versus through a mic, console, and speaker. Our example direct path is 32.8ft. The mic is placed 3.3ft. from the violin and the speaker is 13.1ft. from the listener (a total path of 16.4ft.). If the mic and speaker are flat, then the amplified sound will not match the direct sound. This might seem strange at first but remember that the direct sound has 32.8ft. of HF air loss and LF room reflections, while the amplified path has 16.4ft. The amplified path moves the sonic depth perspective to a range closer than the 32.8ft. direct path. The question is, “How much closer?”
Let’s look at this in two parts: HF (air loss) and LF (room reflection addition). The air loss portion is simple—more air equals more loss, so the flat mic/ console/speaker chain delivers an HF sonic range of 16.4ft. The LF portion is more complex because the directionality of the mic and speaker play a role. Increased directionality reduces the reflections going into the mic and being added to the speaker transmission. Directionality reduces the upward LF tilt, thereby shrinking the image range to even less than the 16.4ft. of actual path. An omnidirectional natural source reinforced with highly directional mics and speakers will experience the highest amount of depth reduction.
Figure 3: Increasing distances between a source and listener modify the direct and reflected sounds with increasing tilt and lowered direct/reverberant ratio. These two factors clue us to range. A highly directional loudspeaker can maintain higher direct/reflected ratios and beequalized to reduce tilt and modify perspective. Click here to see a larger image.
We can make the amplified path match the direct path with equalization. We can reduce the LF addition and restore the HF losses until the direct and amplified paths match. We are tilting the 16.4ft. amplified path in favor of the lows to mimic the 32.8ft. direct path. Now we have a 32.8ft. sonic image depth and the ability to add reinforcement gain as desired. If we have tuned our system to have a matched frequency response over the whole room then we have approximately placed all listeners at the 32.8ft. sonic depth. The sonic perspective can be made closer by reducing the tilt, or pushed deeper by making the tilt steeper.
On-stage, we use close miking to isolate the channels for control. This is a very close sonic image perspective. Returning the sonic perspective to something normal requires adding tilt into the response of the mic, console, or speakers. Mics are all over the stage at different distances and axial orientations, and there are also perspective/free direct feeds (e.g. keyboard or bass), so the speakers tend to carry most of the sonic range load.
The Target Curve
What’s the optimal sonic depth? That’s the million-dollar question, of course, because the answer gives us a target curve for equalization. In practice, we are likely to encounter huge differences in visual depth between performers and near and far listeners. A midsize concert might have listeners ranging from 13.1ft. to 131.2ft., and easily 262.5ft. for an arena. We want to reduce the sonic perspective and bring the audience closer to the performers. But how close? A 50-percent reduction? Bring everybody on-stage? Inside the kick drum?
Figure 4: Progressive doublings of speaker array quantities (1,2,4 and 8). The response tilts with quantity. Equalization can beapplied to reverse the tilt to obtain the desired perspective. Click here to see a larger image.
If we close-mic the instruments and make the system flat from DC to light, we are reducing the depth to an uncomfortable and unnatural perspective. I love Bonnie Raitt but I don’t want her singing directly into my ear like it’s a handheld. If plausibility is important (such as musical theatre), then we need to limit the reduction to something short of detectibly unnatural. Shrinking the distance in half may be enough to bring people in without awareness of the speakers. As the venue scale rises, the amount of tilt required to maintain a median sonic distance goes up.
If we can maintain a constant frequency response throughout the room, we are a long way toward a scalable and consistent sonic perspective. A uniform spectrum keeps the perceived depth constant and the secondary mechanism, direct/reverberant ratio now moves into play. As we move closer to the sound system (and stage) the D/R ratio should rise, which correlates to reduced depth. In the rear, the D/R will naturally fall and the depth increases. Let’s tune the system to create the desired target curve at the midpoint depth of the room and chart the sonic depth variance from there. As we move forward, the D/R ratio rises (sonically closer) and the MF/HF response stays the same (sonically constant). The LF will probably rise as we get closer, a fact of life in PA world (sonically farther). We have a bit of a stalemate in perspective. As we move back in the hall, we lose D/R ratio (sonically farther), keep the same LF/MF/HF response (sonically constant) until we reach the extreme rear, where the LF rises up (sonically farther). We are moving the image away but at a lower rate than natural sound (because we are keeping the HF frequency response fairly constant).
Everyone has an expectation in their head of target curves, but they are not as simple as following a graph. I have done work with the same band, same mixer, and worked closely to achieve their target curve/sonic depth and stored it in the analyzer memory. A night later in a new city, we tuned the system to the same target curve but did not end up there. The hall was much more reverberant and he wanted less LF range up-tilt. My conclusion: He wanted the same sonic perspective. More tilt, less reverb (hall 1), and less tilt, more reverb (hall 2) came out to the same sonic depth.
None of this holds up over the space unless you have tuned the system for uniform frequency response (which is the real job of system tuning). Once this is set, the process of sonic range setting can be undertaken. The good news is that there are very advanced tools out there for gently shaping the response into compliance with your desired target. I will give it a shot, but if the one I guessed for you doesn’t feel right, then we’ll rock it up (or down) until the desired effect is achieved. I won’t take it personally. We are happy to help the process with our analysis tools. But in the end, bringing the audience to the desired sonic depth is system toning, an artistic decision for the mix engineer.