Why Designing Wireless Speakers and Soundbars Is So Different . . . and So Much Harder

June 2014

Brent ButterworthSomeone from a company that sells mass-market audio/video gear called me the other day to ask if I knew anyone who could replace their audio product manager, who’d recently resigned. Right off the bat, I couldn’t think of anyone -- and not because I don’t know enough people in audio. It’s because the job of developing audio products has radically changed in the past decade -- and, in my opinion, has gotten much more difficult.

Most of the audio gear we used in, say, 2000 was straightforward stuff: loudspeakers, amplifiers, receivers. The rules for making all those products were defined decades ago. There’s no big mystery to making a good conventional speaker or amp, and in most cases, the design engineer has the budget to do something halfway decent. Even a $200/pair bookshelf speaker will probably have a reasonable crossover network, a fairly stiff enclosure of MDF, and acceptable drivers.

This isn’t the case with the hot audio products of today, such as soundbars, and AirPlay and Bluetooth speakers. Even models costing as little as $200 or $300 must include built-in amplification, some sort of wireless receiver, maybe a remote control, maybe some sort of iOS or Android app, and, probably, a really slick industrial design -- all in addition to the actual speaker elements.

Often, the speaker sections of these products are compromised to hit as low a price point as possible. I hear this all the time from the audio engineers I talk with. You’d be surprised what almost always gets cut first: the crossover.

The engineers always want to include a decent crossover -- maybe one or two capacitors, two chokes, and a resistor -- but sometimes the money just isn’t there, even for parts whose costs probably total less than $1. Often what you’ll find is a lone capacitor in series with the tweeter, to keep that driver from blowing up. With such a primitive crossover, it’s practically impossible to make an audio product that sounds good, because the outputs of the woofer and tweeter end up interfering so much with each other. But because consumers can’t see the crossover, it’s an easy thing to cut -- even though omitting it can ruin the product’s sound.

The engineer does have a workaround, but it’s not a great one. Most of these products are powered by monolithic amplifiers: class-AB or class-D amps built into a chip. Many of these chips have built-in digital signal processing (DSP) that can be used to fine-tune frequency response, limit distortion, etc. Even a lot of $50 Bluetooth speakers have DSP. Thus, some of the acoustical flaws in the device can be tuned out.

However, there are two complications. First, there’s nothing you can do to correct many acoustical problems. For example, DSP can’t eliminate acoustical interference between a tweeter and a woofer unless the drivers are separately amplified and separately DSP’d. Likewise, there’s little or nothing DSP can do to counteract nasty vibrations in lightweight plastic speaker enclosures.

Second, the DSP included in these products often lacks the power to do much. We think of DSP as something all-powerful, something that can correct any audio problem -- and with the DSP in A/V receivers, it almost can. But with the rudimentary DSP included in inexpensive speakers, an engineer might have only two parametric filters to work with -- and that’s it. He or she decides what the very worst problems are, and fixes those.

But even that isn’t so simple. It requires a solid understanding of psychoacoustics -- you have to know how limiting the bass output might affect the subjectively perceived treble response, how changing the midrange response affects the openness of the sound, etc. You have to know what problems people will notice and what problems they won’t. You have to understand how a tweak to fix one problem might create a new problem. And you can learn only so much about this process from designing traditional speakers, where the flaws and their degree of seriousness might literally be an order of magnitude fewer than in, say, a $200 Bluetooth speaker.

Yet another complication is that most of these kinds of audio products aren’t built by the companies whose brands they wear. They’re built by original design manufacturers (ODMs) in China, Taiwan, and other countries where labor costs are much lower and there’s an established supply chain for electronics manufacturing. Some products built by ODMs are actually conceived and designed by the ODM and submitted to the branding company, which is then allowed to make a few tweaks. Others are conceived and, to some degree, designed by the branding company, with the ODM doing the industrial engineering and the actual manufacturing. Either way, long gone are the days when an engineer or product manager from a mass-market audio company could just walk a few steps to consult with someone on the production line. Now, they have to communicate their needs and instructions through a “language filter,” while watching the manufacturer closely to ensure that any changes made in production won’t compromise performance.

Today, the most successful product managers in mass-market audio are the people who have the political savvy to be able to talk the rest of the team into spending a few cents more on one or two more parts. They’re the people who know how to tune the DSP to get the very best out of their inherently compromised products -- and who have the clout and determination to make sure their tuning isn’t ignored or altered by the ODM. They’re the people who are hip to the very latest trends in technology, and who press to get them into their products before everyone else does -- or, at least, at the same time everyone else does. And they’re the people who are wise enough to judge when design should come first, and when performance should come first.

It’s a tough gig. That’s why I couldn’t think of anyone to recommend. All of them are already employed!

. . . Brent Butterworth