Study Hall

Supported By


Prevailing Trends: Diving Into Modern Loudspeaker Development

Insights from design engineer Sam Feine on some of the elements and advantages in a modern loudspeaker development workflow.

Sam Feine is a loudspeaker design engineer who has worked with multiple major professional loudspeaker manufacturers and as well as in the consumer audio market. I recently caught up with him to discuss some of the elements and advantages in a modern loudspeaker development workflow.

Michael Lawrence: Sam, in your work with multiple loudspeaker manufacturers, what’s your process of creating tunings for new loudspeakers before they go on the market? Also, what are some of the more recent advances in that process?

Loudspeaker design engineer Sam Feine.

Sam Feine: In terms of true tuning techniques, one tool that’s really powerful is FIR (finite impulse response) filtering. A lot of people hear “FIR” and they think “linear phase EQ,” and that’s one thing it can do, for sure, but it’s useful for more than just that.

One reason FIR is so powerful is that, once you make an FIR filter, basically everyone’s FIR convolution gives the same results. It’s not like different manufacturers having different definitions of what Q means, which is a common issue with IIR (infinite impulse response) filters. If a manufacturer puts out an FIR EQ preset, it means the same thing on every brand of DSP you load it into.

Now, the problem is that FIR generally doesn’t work well at low frequencies, at least with relatively short filter lengths. So you’re typically going to have to augment it with IIR filters anyway in the low end. What some manufacturers do is release a library of [IIR] presets for a ton of different DSP platforms, because the filter shapes vary a bit between platforms. Creating all those presets in the different software tools and adjusting the IIR coefficients so they match between platforms is a ton of work, and very time consuming. There is some software that makes this easier – the Eclipse Audio FIR Designer software is great and I recommend it highly – it has different IIR definitions built into it, and it can also convert between definitions for different processors.

ML: We’re primarily talking about filter Q here, right? That different manufacturers tend to define the filter bandwidth in different ways?

SF: Yeah, you could say the manufacturers use different definitions for what a parametric filter should be. And yes, when you do a parametric filter, Q is generally the part most open to interpretation because hopefully you’re not misinterpreting center frequency and gain [laughs].

ML: We’re seeing a “trickle down” trend here – even in the small format powered loudspeaker market, lots of boxes have FIR built in. It seems like we’ve hit a point in the market where it’s become cost effective to build that into even your basic powered speakers now.

One of the boxes I work with regularly uses onboard FIR to linearize the phase response down to about 300 Hz. The physics of FIR dictate that if we want to go lower we have to pay for that in delay, and it seems our industry has pretty much found that saddle point and said, “well three or four milliseconds, we’re good with that.” But if we want to go down to the bottom of the sub range, we can’t wait around for eighty or ninety milliseconds, that’s next week basically.

And I also feel – when you get down into the bottom octaves, once you put that loudspeaker in a room, that room has a huge effect on the response anyway. So up in the high-frequency range, where we have directivity, the benefits of FIR become really potent.

SF: I agree with you completely. I think where it does matter at the lower frequencies is when you’re trying to pair loudspeakers together, or pairing loudspeakers with subwoofers, which in a room – the room has more of an impact down there, like you said. If you tailor the boxes so they have similar phase responses to each other, things are going to work better. If you have loudspeakers from different manufacturers and they’ve done wildly different things at LF, that’s the region where summing really happens (compared to HF where it’s more of an “RMS” or “power” sum).

ML: Is it fair to say that manufacturers are paying more attention to that, making sure that multiple products from the product line are going to be more compatible with each other?

SF: I think that’s fair. And another factor is that I think a lot of manufacturers have paid attention to it for years, but they now have the tools to really make it happen without compromising so much. Every manufacturer I’ve worked with pays a lot of attention within a specific product line, and on top of that, some manufacturers really work hard to make sure that you can cross between product lines as well. That’s a little bit more rare and is sometimes hard to do because different product lines might be designed to do different things.

ML: The other cool aspect to this – that I’ve seen you demonstrate this – is if you’re tuning a loudspeaker in the field, you might put in two, three, four filters. But if you’re building a DSP to go onboard a loudspeaker, or in an amplifier preset, you can build a single filter that creates the entire response that’s needed.

SF: Exactly. And it’s very powerful for a manufacturer to be able to do that. During the development process, when we’re prototyping and listening and tuning, we will typically just stack up traditional IIR filters on top of the “lab curve-based” FIR. We use IIR filters as we’re tuning the loudspeaker because they’re human readable. You can look at the curve and tell which filters are doing what. But at the end of the day, for the final product that goes to the customer, we build those manually stacked IIRs right into the FIR and convolve them together to get one single filter. That’s what gets built into the box or released as a preset.

ML: So in a sense what we’re talking about is an under-the-hood thing that the end user is not involved with. But let’s talk about crossing over that line a little bit – other tools that manufacturers do offer directly to the customer, like prediction software. It may be a stand-alone prediction program or it may be the library files to use in a third-party prediction program. Either way you’re enabling the user to work with your products and see how they’re doing to interact with each other and that is – again – not just a luxury anymore.

Pretty much all the major manufacturers now offer some form of prediction tool, and this is something that has really revolutionized the way we deploy loudspeakers in the field because we can go into the venue and say, here’s my trim height, here are my angles, and if the loudspeakers are put up all right, you can expect them to behave very much like the prediction said they would.

SF: The new tools are fantastic, very accurate and generally easy to use. I also think it’s important to say that sometimes people aren’t comfortable doing the modeling themselves, but you can call up pretty much any major manufacturer and talk to them about it. Every pro audio loudspeaker company I’ve worked with has an applications team. And they will answer your questions about the loudspeakers, they can recommend loudspeakers for your venue, and sometimes they’ll even put it all into the prediction software for you as well. Every manufacturer is different in how they handle that, but they’re generally really helpful.

Study Hall Top Stories

Supported By

Celebrating over 50 years of audio excellence worldwide, Audio-Technica is a leading innovator in transducer technology, renowned for the design and manufacture of microphones, wireless microphones, headphones, mixers, and electronics for the audio industry.

Dante Domain Manager Expert Series on Networked AV (Part 1) - Download Now!