November 2025

Algorithms, Systems, And Ubiquitous DSP
AI makes audio signal processing more pervasive.
Editorial By J. Martins

  Signal Processing—mostly in the digital domain (DSP), but also not forgetting any form of analog signal processing at some stage—is currently an essential discipline in audio. It's always curious to see discussions among hifi enthusiasts about what they recognize as being "acceptable" to do or not do with DSP—when in fact it is virtually everywhere in today's audio technology. And why would "analog" signal processing be any "better" than "digital signal processing" these days?

More intriguing is debating the merits of fully integrated speaker design with optimized amplification and signal processing, which these days is a standard in professional studio monitors ...independently of whether the electronics reside inside the cabinet or not. Why would a speaker design for home listening be any different? As demonstrated countless times, signal processing can be many different things, from helping achieve ideal active crossovers, compensate for the room acoustics, or adjust frequency response targets with different goals (a more enjoyable listening experience being an acceptable objective that is completely different from achieving a linear neutral response for reference monitoring).

The fast evolution of signal processing is what powers today's modern technology. Innovations demonstrated in products we use every day, making possible that we are understood during a call made in the worst possible acoustic conditions, removing environmental noise in our earbuds, making a car ride much more enjoyable, or simply making a $50 portable speaker sound "good" (because it's what millions of people can afford). On the professional audio level, we wouldn't even be able to dream of the sophisticated immersive audio installations with multiple arrays we enjoy today, some of which can transform the acoustics of the whole auditorium at the press of a button, without all the tools available in today's DSP arsenal.

Many DSP companies in the audio industry are systems-oriented organizations that deploy and optimize audio signal processing technologies in products. They may have started with algorithms and signal chains, but they realized the business requires a system design approach. In contrast, many DSP specialists remain invisible to the industry since their efforts are licensed and integrated by others. Signal processing chains are rarely "owned," and are the result of coding for a target device and optimizing for products and results. Optimization often happens in the hands of those working in deployments in specific fields.

Because I attend different audio trade shows in multiple, different industries—mostly unaware of each other—I meet many different companies working in signal processing exactly in the same field, just for different applications. Some doing very well in a market that finds their solutions life-saving, while others struggle to generate awareness in a market that doesn't "buy" DSP. Some making millions licensing algorithms, while others with brilliant code are selling $14 plugins (50% off just for this week).

And all that is also quickly changing. More and more companies are now leveraging artificial intelligence, machine learning, trained models, and neural processors for audio processing, mostly unaware of the work of others. They don't even know who their real competitors are, unless they become visible in the same exact industry and application field (and sometimes because they have a booth in the same hall).

But they all seem to have their targets set on all the "impossible missions" of all the things in audio that were previously unthinkable... (sometimes unethical), and now seem to be the new utopia that everyone's pursuing. Source separation, scene classification, speech enhancement and synthesis, voice recognition, simultaneous translation. All happening.

Audio signal processing is probably one of the fields where we will see a more significant impact from AI. Creating completely new categories of products (such as hearables), creating new markets, while disrupting others. But what I know is that AI will just make audio signal processing even more pervasive. It will be everywhere and no one will even think about it.

Certainly not the users.

J. Martins
Editor-in-Chief

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