March 2026

Simply Better
When we use AI to improve audio signals, it really delivers.
Editorial By J. Martins

  Writing for this issue of audioXpress always happens when we are immersed in the incredible experience that is the most influential consumer electronics trade show in the world: CES.

These days, CES is mostly about the consumer technology supply chain—technology providers and parts suppliers pushing the ODMs and the OEMs to embrace an idea and showcase the best and most cost-effective way to implement it on a real product and the fastest way to bring it to market successfully.

When it comes to audio technology, CES allows seeing the most extreme of the equation, from sets of reference designs from where to start, all the way to projects in-the-works, with engineering and technology partners involved in a demonstration. As I often explain in my CES reports, most of the demos I attend are not of working products, but of loose parts of the technology ecosystem required to enable new product concepts and categories. And that can go from wireless speakers to home cinema systems, from gaming equipment to new headphones or hearables with cutting-edge features. These days, it's increasingly about running small, optimized models on ultra-efficient devices to enable new possibilities in voice command and control, or new generations of devices that can handle speech recognition for transcription and translation. All things where audio technologies are at the core and spreading to new product categories and applications.

That's what happened at CES 2026, and we are already seeing it multiplied in new product announcements. And note that I have avoided using the term "AI", even though the motto for CES 2026 could very well have been "AI in Everything We Do," as one manufacturer proudly displayed. "Artificial Intelligence" was the common denominator for the most interesting announcements and demonstrations we have witnessed this year. At CES 2026, one way or another, all the key AI technology companies were present and extremely active. And the real innovative aspect of CES this year was to see the many different ways in which companies are implementing AI in very different products, large and small, connected or not.

Interestingly, running optimized trained models on-device (Edge AI is not a "thing" in the consumer segment) was the most common approach at CES. As AI was increasingly used to serve function, it doesn't matter how large or small, sophisticated or simple a product is, AI is now embedded and optimized to run independently and in the most efficient way. AI is used to remove noise on audio signals, to separate multiple sources, to detect intent, to improve communication, and to simplify users' interactions.

That can be in a car, or an appliance. Product designers and many technology companies are placing AI as close as possible to the sensor or source of data, where it can be effective and seamless, while at the same time, there are new exciting product categories that remain connected when it matters: to run a sophisticated room calibration on a soundbar, or to manage the user experience across devices, including playing audio in the kitchen and moving to headphones in the street or to an autonomous taxi—now a commercial reality in Vegas and in major US cities.

And as these new "intelligent" features are moving from being the magic "AI" acronym that still sells (for now), to become real-time inferencing and decision-making applications embedded on products, the possibilities for expanding audio applications are also growing faster than ever. The challenge for product developers is not when, but how fast and efficiently they can evolve their product design process to leverage "AI in everything." Including starting to use it to accelerate system development, for simulation validation, to better understand signals, and even to measure those signals more efficiently. The test and measurement companies are embracing these possibilities as fast as they can, including to process larger sets of data in production and quality control.

There's a lot of talk about what AI can or cannot yet do, and about what is and isn't. But when we use it to improve audio signals, it really delivers. So no one should be surprised to see it embraced on the test bench, in consumer or high-end products.

J. Martins
Editor-in-Chief

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