Why use automatic EQ?
Automatic equalisers are not foolproof, writes Anthony Grimani. Are you risking your reputation by using one?
I talk to lots of integrators who promote integrity as a core principle of their businesses and make promises to satisfy and please their clients. I don’t think this is just propaganda coming from them; they really mean it and intend to provide the best product and service available.
Unfortunately, they may not realise that they are letting themselves and their clients down when it comes to delivering the best possible sound quality from their home cinemas. Yikes! Here’s why.
The final step of a home cinema project is to calibrate and tune the sound and picture systems. To do this well for the picture is pretty straightforward these days – a lot of displays even offer pretty accurate automatic calibration systems.
The same concept is available for sound systems.
Unfortunately, the results typically aren’t quite as rosy. Audio calibration, especially selecting target curves and tuning EQ, is a really complex process that we aren’t yet consistently able to control.
I know it’s not cool in 2019 to admit that there are fields where we lack omniscience, but this is one of them. It reminds me of the recent comic book movie Venom, which is about an alien symbiote (Venom) who has to bond with a human in order to survive on earth and provides its human host a bunch of nifty upgrades in return – such as increased speed, strength and resilience. Unfortunately, it takes a while for Venom to find a compatible host. The best minds in the world try to force the bonding process, which they don’t understand, and fail spectacularly. Similarly, sound systems and rooms must exist in a symbiotic relationship in order to produce great sound, and we don’t yet fully know how to ‘help’ this to happen.
Among the many variables are the effects of speaker directivity across the entire frequency range, the room sound reflection decay character across the entire frequency range, and the effects of room size on perception of direct sound versus early reflections, mid reflections and late reflections. It takes a trained audio/acoustical engineer with years of experience and high-tech gadgets to account for all these variables – and the process still involves a lot of trial and error! Listening tests and final adjustments based on audible performance are still the ultimate arbiter.
Now I’ve got a question for you. Do you think an automatic system can do all of that? Just think about it. How much AI and processing power would it take? That’s not available in an AVR. Or even a surround processor, really. For the most part, we’re talking about simple microprocessors, not supercomputers. I’m really not hating on automatic systems, here. They can do a decent job as far as they go – especially the human-directed ones with the ability to control target curves, impulse response window sizes, etc. But it’s a fact of life that they lack the sophistication to deliver the best symbiosis between the sound system and room.
Designing these automatic systems is an admirable pursuit – and I think we should stop and applaud the people working on them. They have a great challenge, and I hope they continue working to make them better and more sophisticated to the point where they truly consider every variable.
OK, so I can say all of this, but that doesn’t say you believe it. You’re skeptical; I get it. Maybe you’ve done a few rooms yourself, using an automatic room EQ correction system, and think it sounds pretty good. That may be true; you may have been lucky and stumbled on a few scenarios where an automatic system worked OK – it does happen. Not every time, not predictably, but it does happen. But how do you tell when things have gone wrong? What are the most common issues and how do they manifest? What do they sound like?
- The overall spectral balance may be wrong – perhaps just slightly wrong, but definitely not natural. The bass may be shelved up or treble shelved down, resulting in heavy, bloated or muddy sound. Conversely, the treble may be shelved up or bass shelved down, leading to thin, lean, wimpy sound. This is typically the result of the system using the wrong target curve, the system’s inability to accurately measure and correct to the chosen target curve, or inconsistencies in the overall sound power.
- The midrange may be too aggressive or too laid back. The sound smacks you in the face yet lacks sparkle or true impact, or it’s hard to hear midrange details. A great way to evaluate this is to simply listen to dialogue. With an aggressive midrange, you’ll hear every word but voices will be oddly dull and nasally. If the midrange is depressed, voices will sound full and strong with lots of crisp sibilants, yet you somehow struggle to understand words.
- Absolute and relative levels are off. Wrong absolute level means the system plays too loud/soft based on where the volume control is set – not a huge issue, but it may lead to clients complaining that they have to turn the system ‘way up’ to hear, or that it’s louder than their other systems when they use the same volume ‘number’ (darn digital volume displays). Wrong relative level means some speakers are too loud compared to others. This can also be a somewhat deceptive problem to ‘hear’ unless the speakers are way off by multiple dB.
- Distances/delays are off. Surprisingly, even small errors in distance compensation delay settings (0.1ms) can shift the entire immersive sound field left/right/front/back to an alarming degree, creating the sensation that the left side or right side of the room is way louder – or even that most of the sound is shifted behind the listener. Individual effects also come from the wrong position relative to the screen and 3D space. Just sit in the center seat and listen to a mix with a lot of ambient space. Does it sound like one part of the room around you is ‘louder’? Probably a delay issue.
- Polarity of one or more speakers is wrong. You can avoid this by wiring everything correctly in the first place, but mistakes happen. If an automatic system fails to detect it (either because it doesn’t detect polarity at all or it does and gets it wrong), there will be a large hole somewhere in the soundfield. Or you’ll get no imaging between the offending speaker and its companions.
- The subwoofer crossover is set too high/low. This is a topic that should be its own column. Choosing the right crossover point requires a lot of information about the speakers’ bass extension and output, coupled with the room’s standing wave character. Is it possible for an automatic system to input all this information and crunch the numbers correctly? Yes, theoretically, but… they don’t. Unfortunately, stuff can get serious. A crossover that’s set too low can damage bass drivers in the main speakers. A crossover that’s too high has the less-dangerous-but-still-annoying effect of bass displacement (you localise stuff to the subwoofers that you should hear from the main speakers).
So, now I’ve thoroughly ruined automatic systems for you forever…
No, but seriously, please be careful how you use these things, because they often do go astray in one way or another. Ideally, everyone would have the time and money to hire a good human calibrator with the best test gear, to budget three days and to tune equalisation by hand for all 19 (or however many speakers/subwoofers) in a full immersive system. Yes, I really do wish everyone did this. But I recognise that they don’t, sadly.
At the very least, check what your automatic calibration system did.
- Listen to The 5.1 Audio Toolkit disc (it’s a thing, look it up) pink noise tracks and compare the sound system to reference earphones like Etymotics to verify that the spectral balance is close. Try different target curves, if available. If the automatic system fails to even get in the ball park, try running the whole process again – perhaps with mics in a different arrangement. One calibration can sound terrible, but the next one is pretty decent. Save your first result, though, because they next one could always be worse… Sigh. Why are these things so inconsistent?
- Use an SPL meter to verify levels. The trick here is to use a known source for the test noise and play it through the system with EQ. (The internal noise probably bypasses EQ.) I know the Toolkit is right, but I can’t speak for other sources. However you do it, make sure your source files are accurate relative to each other and to reference level, then bump channel levels up or down as necessary. Note that I do tend to give automatic systems some leniency on relative levels due to differing methodologies for measuring.
- Check distances/delays with half-point test signals. These are identical narrowband noise signals played from pairs of speakers (left/centre, left/right, right/side right, etc.) The test noise should sound like it is suspended in space between the speakers (imaging), not coming from one/both speakers individually. The only source I’m aware of for signals in this precise form is the Toolkit; however, it’s 5.1 with EX, so what you get is discrete L/C/R, front to side, and matrix side to back (no height/top). That takes care of most of the key imaging points, so it works out. Click delays up or down to centre the noise between speakers. Careful! One click may be all it takes…
- Get a polarity tester. Quick and dirty. Play noise pulse, hold tester up to speaker. Green light, done. Actually, it’s slightly more complicated, because some speakers are designed with certain drivers reverse polarity (red light). Just use your common sense to figure it out, or ask the speaker designer to clarify.
- For crossovers, yeah… You really need to analyse standing wave patterns and measure speaker bass extension versus output. I’m guessing people mostly don’t want to do that. At least check the speaker documentation to make sure the crossover is above the speaker’s rated -3dB point for bass. That’s probably not going to get you the smoothest bass response (unless you’re super lucky), and it may not even be enough to protect the speaker from damage. But you have done something to protect it.
I’m looking forward to the day when we fully understand how to create a symbiotic relationship between sound systems and rooms, and we can make it all happen by pushing a button. I want that day to be here. I really do. But we’ve got a while before we get there.
In the meantime, I just want to make sure you don’t run into a scenario where the client questions your integrity and your promise of satisfaction because an automatic calibration system let you down. It’s fine if you didn’t realise this was a thing. I run into folks all the time who legitimately think these systems are 100% foolproof and feel totally fine guaranteeing their results to clients. If you’re good with that, and they’re good with that, then I’m good with it. But I still think you can do better!