Ensuring clarity in an audio system
Clarity is the single most important aspect of audio system design, writes Anthony Grimani. Are you doing it right?
What’s the most important thing about a home cinema audio system? Bass? Dynamic range? Stereo separation? Surround effects?
Nope – it’s none of those.
I hate to get mundane, because all that stuff’s really exciting. The most important thing is really dialogueue clarity. And this is not just true for home cinema audio systems. It’s true for public address, commercial cinema, sound reinforcement – you name it. Ask anybody who works in sound and they’ll say, ‘Yeah, well, of course. If you can’t understand what folks are saying, what does the rest really matter anyway?’
I bring this up because I’m seeing a lot of systems with compromised dialogue clarity due to the dialogue being produced by two or more speakers. Clients are complaining; it’s a bad thing, and I’ll explain why in a minute. But how or when does this happen? What’s an example? Here’s one: a stereo sound bar below the TV in the master bedroom – drivers on both ends of the sound bar producing dialogue. Or, a pair of stereo speakers flanking the screen – no centre speaker, either because no one wanted it or because there wasn’t space for it. Or what if there’s a bug in the audio mixing of the content or the routing in your system that causes dialogue to come from the L/R speakers even though there is a physical centre speaker? All of these happen way more often than they should. Let’s dive in and see if we can sort out this mess.
OK, so why is it bad to have the dialogue coming from two different speakers? The short answer is that our ears/brains add the two sound sources together. If we get the same sound of a person talking from different sources (speakers), our brains don’t add right and we end up hearing mushy, hollow, vacuous sound with no clarity or crispness. The official geek term for this is ‘comb filtering’. There, now you have a new term to impress your clients. Don’t say I never gave you anything! The long answer gets way more complicated and deals with all sorts of science-y stuff like arrival time, constructive and destructive interference, horizontal vs. vertical polar response, etc. If you want to know more about all that, I’ll tell you. But not here. Hit up my e-mail address, noted at the end of the article.
Great, now I’ve blown up all your systems. How do we fix this? Let’s start with the simple bedroom sound bar. The core problem is that the sound bar is two channel stereo but dialogue is mono. So you have the dialogue coming out of both the left and right part of the sound bar and adding together in different, weird ways for people on the bed. There are a couple of solutions, here. The first is to use a (gasp!) mono sound bar. Yeah, I know it may not be as ‘spacious’ or ‘enveloping’ as a stereo sound bar, but how much are the clients really going to use it for spacious music as opposed to dialogue on TV shows and news broadcasts? Boring realties, here, people. Another solution is to use a true LCR sound bar with separate drivers for the Left, Centre, and Right. Mono dialogue gets routed to the centre part, but you still have the left and right driver groups for stereo separation – what little is possible with speakers that close together for a listener way across the room. But there’s a little ‘gotcha’ even with mono and LCR sound bars. Make sure the centre drivers are not the horizontal mid-tweeter-mid (M-T-M) array that is so common. This is a bad configuration. Can you guess why? Look back at what we just said about the same sound arriving at your ears from different sources. Those two woofers separated by that tweeter are two different sources! Now, there are things that can be done in design to mitigate the interference patterns from the two drivers – such as using a tweeter that plays down to a much lower frequency than the typical 2-3 kHz, or smushing the drivers as close together as possible. But this type of effort in design is far from the norm. You can search it out. For example, I found one the other day in the Samsung HW K950. I’m not suggesting that all Samsung sound bars are made this way so do a bit of research! But do take note of the design of the centre element and look for something similar in the sound bars you use. Last but not least, take advantage of the companies that will custom build your sound bar. Tell them you don’t want an M-T-M (Mid-Tweet-Mid) Centre array. Instead, you want M-M-T (orientation frequently used for Left/Right arrays in sound bars). It’s still not perfect, but it’s better than M-T-M.
And now what about those stereo speakers flanking the screen in applications where a Centre speaker can’t or won’t do? First, try really hard to use a Centre speaker. If you’ve exhausted every possible option for a physical speaker, choose the speakers you’re going to use for phantom centre very carefully. The first thing to look for is flat midrange frequency response, especially around the middle frequencies. A lot of designers intentionally put a midrange dip in the response to make a speaker sound more ‘laid back’. All this really means is that the speaker is missing critical midrange detail. That same detail gives dialogue its clarity. Second, make sure the speaker has really wide, uniform horizontal dispersion. It needs to be broadband, not limited to some narrow frequency range. So how do you tell these things about the speaker? You need detailed axial and polar frequency response plots. Not a lot of manufacturers provide those, so you are going to have to press them or maybe even switch to a brand that actually has that data for their speakers. Of course, don’t ignore the room. Echoes/sound reflections really tend to kill intelligibility. If the room is right on the edge, switching from a physical centre speaker to a phantom centre can push the sound into mushy-land. Go with a little more absorption in the room if you have to use a phantom centre. Finally, equalize! The response of the two speakers making the phantom centre must match precisely. If all else fails, you need an EQ on the two speakers in order to accomplish that. I won’t get into the reasons why the response might not match, but ultimately you may be forced to use electronic correction.
Last but not least, what the heck can you do if the mix is funky (like dialogue mixed into only the L/R or all three LCR channels), or your source device is determined to output stereo mixes from the L/R speakers without giving you the option to engage matrix decoding in the surround processor and use your Centre speaker? This happens a lot with cable/satellite boxes (particularly for local affiliates with Kangaroos mixing their sound) or gaming console/computeresque boxes with internal audio mix engines that want to keep a 5.1 or 7.1 pipeline open all the time. Well, you’re going to have to create an alternate audio path for the compromised content. Usually, this takes the form of downmixing it to a 2.0 output. You have to keep the HDMI multi-channel path open for 5.1, 7.1, and immersive surround, though. The trick is to use a little device called an HDMI audio de-embedder to lift the audio out of the HDMI path and downmix it to a 2.0 analog output (RCA patch cords). Bonus. This devices doesn’t remove the audio from HDMI, so that pipe stays open. With access to both a bitstream/multi-channel feed from HDMI and a 2-channel downmix through analog, you can set up the surround processor and automation system to allow the client to switch to the backup when dialogue sounds wrong in the multi-channel feed. Yes, this is going to take a little creative programming and client education, but it’s nice to have a fallback instead of fielding complaints about not being able to understand the news anchor or the play-by-play guy.
Whether you realize it or not, there are a whole lot of ways in which your clients can end up listening to unintelligible dialogue coming from two speakers. It can happen out of ignorance. It can happen as a result of misguided priorities. It can happen as a result of poorly produced content. Regardless, the task falls to you to sort this out and make sure dialogue is crisp, clear, well-balanced, and easy to understand. It’s really the most important thing you can do as an audio integrator. Are we clear?
Chase Walton contributed to this column.