Is the film industry's X curve suitable for home theater?
System Calibration vs. Program Compensation: Will a house curve deliver sonic bliss? Why you need a house curve
It was a great day for me back in 1996 when I took delivery of the high-end AudioControl 1/3-octave equalizers I had been lusting after for a number of years. At long last I was moving up to the equalizer "big leagues!" At the time I had been installing pro-audio sound systems long enough to know that you don’t set these equalizers by ear, so delivered along with them was an AudioControl real time analyzer/pink noise generator.
It helps to be completely immune to pink noise if you aspire to fine tune your system with sophisticated equalizers (a dubious benefit of installing pro systems for any length of time). My poor wife, I think she ran out of the house screaming and pulling her hair out, but after an hour or so I had dialed in a nice, flat response curve. I popped in a CD and sat back to enjoy the wonders of a perfectly tuned system. But horror of horrors, it sounded awful! It was all thin and shrill-sounding with virtually no bass. Where had I gone wrong?
If I had bothered to consult the AudioControl manual before I fired up the pink noise, I would have known I was headed for disaster. Unfortunately, the manual offered no marital advice, but buried in the back pages I found this passage: "When you have flattened out the analyzer’s display curve as much as possible, you’re done with this portion of the adjustments. But you’re nowhere near finished. When you play music through your system at this point, you might not be completely pleased with the result. With the equalizer you have compensated for all sorts of factors, but you have also run up against an important psychoacoustic effect: Flat response doesn’t sound all that great to most people."
Boy, no kidding, I was thinking. The manual went on: "Through long experience, acoustic experts have determined that a ‘curve’ which is tipped towards more low frequencies is much more desirable for music reproduction. It’s sometimes (although not totally accurately) referred to as a ‘house curve,’ a term that comes from the fact that adjusting music systems in individual theaters and auditoriums to achieve the most ‘listenable’ sound tends to produce curves like this, with a significant reduction in higher frequencies. Figure 14 shows a ‘room’ curve which can serve as a starting point for your secondary equalizer adjustments."
Here’s the Figure 14 from the AudioControl manual, a 1/3-octave frequency response graph:
So there you have it: If you want your system to be "listenable" – i.e., sound balanced - you need a house curve.
Flash-forward a few years to 1999, when I got my first computer and started participating in the various home theater forums: At the time enthusiasts were just getting into using parametric equalization to flatten their subwoofer’s frequency response. I noticed people were saying they weren’t happy with the results, claiming their subs had gone from flabby and boomy to thin and insubstantial. Recalling my experiences from a few years earlier, I recommended that these folks try a shelved house curve instead of flat response, and most seemed happy with the results. The Internet was a big place even then, so I’m not sure I can lay claim to pioneering the concept of the home theater house curve. But at least one salty fellow
blames yours truly for the whole “mania” (even if he isn’t clear on where I got the idea). What exactly is a house curve?
As you can see from the AudioControl manual’s picture and text, a house or room curve actually encompasses full-range response. However, on the various home theater forums the term has come to refer specifically to subwoofers and low frequency response, so that’s what we’ll focus on to start. In the second section we’ll discuss some theory and a full-range house curve.
I’ve looked at the on-line manuals for most of the pro-audio equalizers on the market, and the passage from the AudioControl manual is the best description of a house curve I’ve seen. Yet their definition is not as clear as it could be. Simply put, a house curve is perceived
flat response as opposed to measured
flat response. AudioControl also doesn’t tell us that a house curve is as critical in small rooms (where we use our systems) as it is in the theaters and auditoriums they mentioned. Perhaps even more so, because small rooms need a more aggressive slope than large rooms to sound balanced (we’ll discuss why that is in Part 2).
Since no two rooms are alike, the million-dollar question is: How do you determine how much of a slope you need for your room?
On various home theater forums you will find people recommending all sorts of methods for acheiving a room curve. One of those is what I call a "one size fits all" approach - a 12-dB rise in the subwoofer’s response, or some other rigid figure. Feel free to ignore such advice. Dialing in a house curve is not a "one size fits all" endeavor. The problem is that an ideal house curve varies from one room to the next. The rule of thumb is, the smaller the room, the steeper the slope needs to be.
Yes, that point has already been made. But it bears repeating, because even though all the rooms in your home are small compared to a movie theater or auditorium, the small room/large room rule applies here, too. A bedroom theater verses a system in a family room are a couple of typical small room/large room scenarios in the home. As such, a "set figure" approach won’t work unless you’re lucky enough to need precisely that slope.
Like the "one size fits all" approach, another problematic technique is what I call the "compensation curve method," which relies on performance shortcomings of the Radio Shack SPL meters. It’s a well know fact that the meters (which are often used with frequency-specific sine waves to measure room response) naturally roll off at lower frequencies. In the interest of more accurate SPL readings there are readily available compensation values for the meter’s deficiencies. However, the "compensation curve method" holds that equalizing for a flat curve without the compensation values will automatically get your house curve. That is, you will have to boost your lowest frequencies to get response that measures flat, and when you apply the compensation values you’ll find your lowest frequencies actually have a rise.
The problem with this method is that the SPL meters’ response doesn’t start to roll out until 50 Hz, and then it only falls at a rate of about 4 dB/octave down to 20 Hz. Above 50 Hz the meters are reasonably flat, albeit miscalibrated by a decibel or two. So, what the "compensation curve method" gets you is:
- Essentially a "one size fits all" curve, since it rises at a pre-determined rate.
- A curve that doesn’t rise much.
- A curve that starts too low (as we’ll see later, you probably want your house curve to start rising where your sub rolls in, which is about 80-100 Hz on most receivers).
A "compensation" curve may work well in some really huge rooms – say, a medium-sized church – but not in most home theaters, which typically need quite a bit more than a 4 dB slope that starts at 50 Hz. An easy way to determine the house curve you need
If all this is sounding complicated, it really isn’t. It’s actually fairly easy to determine the house curve your room needs. Here’s a simple method that has worked well for me.
The first thing you need to do is use an equalizer, preferably a parametric model, to get your subwoofer's response reasonably flat. By that I mean eliminate as many peaks and low points as possible. It doesn’t matter if the response you end up with is more-or-less flat-lined or has something of a tilt. The idea is simply to get as smooth a response line as possible. Go for the approach that uses the fewest number of equalizer filters.
After your sub’s response is reasonably smooth, play a couple of sine wave test tones, one at 100 Hz (or whatever your sub’s crossover frequency is), the other at about 30 Hz. (Naturally, you don’t want either of these to be in a null - shift your test tone up or down if you have to.) With response smoothed, the 100 Hz tone will probably sound louder than the 30 Hz tone. If that’s what you find, your sub’s response needs to be adjusted so that both test tones sound like they’re the same volume level.
Yes, that’s a highly subjective evaluation, but remember a house curve is perceived
flat response – that is, it sounds flat, not measures flat. Thus it has
to be subjective.
So, using the upper frequency (100 Hz in this example) for your baseline SPL reading, if you find you have to increase the sub’s volume say, 8 dB to where your lower frequency (30 Hz in this example) sounds as loud, then you need an 8 dB slope between 100 and 30 Hz.
As you can see, since you’re merely trying to get the higher and lower frequency test tones to sound the same, you don’t have to worry about any complicated formulas or calculations. Your room does it all for you automatically - great news for those of us who are mathematically challenged! Dialing in your house curve
If your baseline frequency response (i.e. before you smoothed it to determine the house curve) had exaggerated lows, you could simply re-equalize the sub to show a 8 dB boost between 100 and 30 Hz (keeping with the example above). However, if your base response is such that smooting out the peaks and valleys gets you nearly flat response, the easiest way is probably to apply a shelving filter. Some equalizers have shelving filters, but they’re typically pricey pro audio parametrics. The popular Behringer Feedback Destroyer unfortunately doesn’t have a designated shelving filter. However, our esteemed Administrator brucek came up with an innovative way to effectively create one. After equalizing his subwoofer he also found a need for a shelving filter to precisely dial in the slope he needed. Here’s what he wrote about it a few years ago at another popular forum:
"I had an idea, that if I added a single filter up around 400Hz with a very wide bandwidth and a large cut, designed so that its final effect would reach down to about 30Hz, that I would have a smooth drop from 30Hz all the way to 400Hz. It would essentially tip or hinge the shelf at 30Hz and continue to drop increasingly, all the way to the filter’s center frequency."
If that sounds too technical, perhaps this picture brucek created will explain it better:
You can see from the nomenclature that the filter is very broad: two octaves, centered at 366 Hz, and cut 15dB. Take a look down at the subwoofer range and you can see the effect it would have on a flat response curve: A 4-dB drop between 30 and 100 Hz, with response flat (shelved) below 30 Hz.
Keep in mind that you will have to adjust this house curve filter to suite your particular system and situation – your crossover frequency, the response curve you’ve dialed in, etc. For instance, if your equalized response already has something of a downward tilt, this filter might be ideal for you, to increase say, a 6 dB slope to 10 or 11 dB. However, if your response is closer to flat-lined, you’d probably want more than a 4 dB slope. That could be accomplished by cutting the pictured filter even deeper, but you would end up with response that continues to rise below 30 Hz, not shelving at that point (more on that shortly). To get a steeper slope and retain the 30 Hz shelving, you would need a tighter bandwidth, with the center frequency moved down to a lower frequency.
Thus you might need to tweak the shelving filter until the 100 and 30 Hz test tones sound like they’re at the same volume. When you accomplish that, you’re probably golden. I say "probably" because since every room and system is different, you should let your ear be the final determining factor. (NOTE: You can effect a hard knee curve
by applying a couple of extra filters to the shelving filter. A 36/60 bandwidth filter centered at 56.50 Hz and cut –2 dB, and a 17/60 bandwidth filter centered at 27.60 Hz and boosted +1 dB will flatten the shelving filter’s “rounded” response [bandwidth and center frequency values based on using a Behringer 1124 Feedback Destroyer parametric EQ]. See note below. )
When you're satisfied with the way the 100/30 Hz test tones sound, re-check your equalization between those two points. The idea is to get as straight line as possible between them. To check your equalization, play some reference CD’s you’re familiar with and see how things sound. See if you can find some music where the bass line ranges from very low notes to very high. Ideally they should all sound the same from note to note – no hot notes, no weak notes (assuming the recording is good to begin with). Adjust your slope as you feel is needed. You may need to move the point where the curve starts to a higher or lower frequency. You might not need straight-line response between the starting frequency and 30 Hz; you might need a line with a sag or a hump. Whatever it takes to get those bass lines running smooth from top to bottom, that's what you want.
It’s not a bad idea to take the Fletcher-Munson effect into consideration, in that you want to dial in your house curve at your usual listening volume. If you set it for a higher volume than you normally listen to, it will sound bass-shy when you turn it down. The inverse is true if you house-curve for too low a level. Why shelve response?
Earlier I mentioned shelving your house curve. I personally prefer to shelve response at about 30 Hz rather than allowing the curve to continue rising. What that means is the rise stops at that point, and is flat from there down to the sub’s lowest limit. I initially kept the slope rising all the way to the bottom, but I found there was an overabundance of ultra-low energy. The bass in music sounded "heavy," and with movies things like car doors slamming had all the "umph" of a distant explosion. Certainly did not
sound natural. Shelving response at 30 Hz solved that problem. Since every room and system is different, it doesn’t hurt to experiment, but if you find a continually rising curve sounds bad, try shelving it. Music should be greatly improved, as I’ve found that bass detail and resolution gets obscured if response continues to rise. There’s no penalty with movies, since they have greatly exaggerated extension that essentially overrides any shelving.
So that’s my "quick and easy" house curve method. To be fair, I’ve only used it once, in the house where we used to live (haven’t tweaked my system yet in our new place). However, I’ve been recommending this method to people for a few years now and no one has ever come back and said, "It sucks, what’s your next bright idea?"
So, give it a try and see if works for you. I have found it to be very musical - balanced and accurate with lots of bass detail (assuming your sub is capable of rendering detail) - yet there’s enough impact with movies to vibrate the sofa. Who needs bass shakers?
NOTE: More recently I’ve decided that I greatly
prefer the sound of a hard-knee house curve
, with a sharp decline from the shelving frequency, rather than one derived from a simple shelving filter or re-aligning the crossover curve. Fellow Shack Moderator Ayreonaut’s experimentation led him to prefer to shelve his response lower than 30 Hz. Lately I’ve dropped my shelving down to ~28 Hz, and I like the way it sounds. Here’s Ayreonaut’s thread on the subject: House Curve Options Thanks to brucek for kindly allowing us to use his shelving filter picture.