October 2025

The Appealing Promise Of Tiny MEMS Speakers
Thinking of the analogy between the CV equation and the inherent merits of MEMS.
Editorial By J. Martins

  audioXpress has been reporting on the promise of microelectromechanical systems (MEMS) speakers for nearly 10 years. Mike Klasco first gave a talk about MEMS loudspeakers at the ALMA Symposium and Expo event, preceding CES 2015, and wrote regularly about MEMS speakers' evolution. A decade later, we are seeing some actual progress, and products are available in the market.

But MEMS speaker drivers are far from their original promise of disrupting the industry or even rivaling established technologies. Which raises questions until we dive into the details to understand that there's very little in common among all the MEMS transducers products that have been launched and promoted in the microspeaker class.

The main thing in common is that MEMS typically include "microscale" mechanical structures (such as movable parts) and microelectronics all embedded on a silicon substrate. Other than the foundational technology, multiple MEMS devices share the same manufacturing processes derived from semiconductor fabrication.

As we stand in 2025, the pioneering companies that have entered the industry with "MEMS speakers" have done so with completely different approaches. And some very interesting original concepts have not yet even made it to market.

As Mike Klasco notes in this issue's Market Update, OEMs remain too price-sensitive for the current generation of technology, while fitting a completely new part and corresponding processes into today's well-oiled manufacturing supply chain takes some indisputable benefits or some actual arm-twisting to make it happen.

In this issue, audioXpress provides a comprehensive overview of what is happening in MEMS microspeakers. We hope this will inspire confidence and motivate further investigation into the merits and specificities not only of different vendors' technology, but of the different products from the same company. Yes, to say that not all MEMS speakers are created equal might sound cliché, but it's actually more accurate than with any other microtransducer class.

And talking about transducers, in preparation for this issue, we are proud to recover an interesting text that remained unpublished, authored by Steve Hutt and Roger Shively on "Subjective Evaluation of Production Vehicle Audio System Variance." This is an interesting work that offers multiple approaches to avoid subjective interpretation and help solve discrepancies between "what is expected, how it is conveyed, and how it is confirmed".

Interestingly, that was the exact departure point for an interesting Audio Engineering Society (AES) paper (7530) on "Loudspeaker Production Variance," written by Steve Hutt and Laurie Fincham (THX) and presented at the 2008 AES Convention. In this detailed work on loudspeaker parameters and its characterization, the interesting Coefficient of Variance (CV) is introduced as a practical way to provide "a path to find deviation trends and opportunities to reduce end-of-line variance" of loudspeaker components and processes.

I found this extremely relevant in today's context, particularly in the perspective that "Loudspeaker production variance is vast as compared to other audio electronics"—as stated in the paper's conclusions. I couldn't help but think of the analogy between this important CV equation and the discussions around the inherent merits of MEMS speaker consistency and uniformity that result from its fabrication process. Something that most of the MEMS pioneers believed would stand out as such a paradigm changer that it would trigger a race for adoption—particularly in a class of products such as personal audio and hearing assistance, where the ability to provide personalization benefits from an absolutely consistent technology platform.

The other inherent Unique Value Proposition is that MEMS microspeakers are simply a better transducer with predictable acoustic performance when an audio signal requires intense digital signal processing. MEMS work better with sophisticated DSP algorithms that need accurate and stable transducer behavior for optimal sound reproduction.

I will end with another relevant quote from the "Loudspeaker Production Variance" paper by Steve Hutt and Laurie Fincham: "The vigilant loudspeaker manufacturer will manage production variance by developing in-process procedures that allow them to set tighter tolerance specifications and ultimately achieve improved fit and function consistency. Applying good quality practices can help move the loudspeaker industry away from the esoteric precepts that loudspeaker design and production is an art."

Again, as always. It's what you do with it.

J. Martins
Editor-in-Chief

Subscribe To audioXpress Magazine
Click here to subscribe to audioXpress magazine.