From The Editor's Desk
In a 2014 presentation at the 137th Audio Engineering Society (AES) convention, Paul Beckmann, the founder of DSP Concepts, revealed the results of a search within his LinkedIn network to determine how many people had "Audio Engineer," "Embedded Software," and "Audio DSP" listed in their profiles. He made a Venn diagram of the results, which showed that very few people included "Audio DSP" and even less (11 out of a universe of more than 100,000), actually claimed to do all of these three things. To illustrate the complexities and challenges of current audio product development and particularly in making an embedded audio product, Beckmann also identified skill sets in electrical, mechanical, acoustical, and user interface software development as basic requirements.
It is not surprising that digital signal processing (DSP) continues to be one of the best kept secrets in the audio industry. The people who actually push DSP forward are not many, while the required "math and numbers" demand a lot of intense work to translate into code by a whole lot more people, who simply aren't there.
A growing number of companies in the audio DSP field deserve recognition, even if sometimes they don't necessarily want to be in the limelight. Mostly, because their work is done behind the scenes in the industry, mainly focused on licensing IP. For that reason, it's only natural that they don't exactly publish tutorials explaining how they do what they do.
In professional audio and broadcast applications, leading companies evolved from the days of analog audio, later embracing digital signal processing, while others started with DSP. Still, many rely on selling hardware to package their processing tools, avoiding getting into too much detail about how things work — just offering a few basic knobs for the user to tweak to get the best result.
In contrast, software companies are extremely more visible, currently offering some of the most sophisticated audio processing ever available: removing background noises, replacing notes and whole polyphonic sequences, or doing sound source separation within existing recorded tracks. Of course, those companies don't detail how their algorithms work, and they focus on explaining the user interface for their application or plug-in. Yet, that's where we find today's most vibrant community of audio developers, particularly since the availability of mobile platforms and the boom of the app stores. But again, not many are actually involved with DSP. Most understand how to implement DSP algorithms as long as they are offered in ready-to-use "blocks."
Increasingly, on all product development fronts, companies are combining DSP algorithms with artificial intelligence (AI). That's the vanguard of digital signal processing these days and lots of very clever things are already made possible because of AI.
Room correction is becoming a commodity in integrated home audio amplifiers, even if combining calculations performed remotely, in the cloud, and stored for active processing in the local device. Same with headphones and head-related transfer functions (HRTF) used for personalization. But soon, all that digital signal processing and AI will be able to happen directly on-device. Even in portable and wearable devices.
Our current mobile platforms are now becoming capable of running those extremely intense computational tasks for real time processing. That's precisely why consumers are now able to benefit from digital signal processing being used so extensively. And thanks to AI, users no longer need to learn control interfaces for things they don't need to know are running in the background.
Still, no matter how powerful the processing and number of sensors, more and more, we need "audio engineers" who are able to understand how signal processing algorithms benefit our experiences, and provide algorithms to feed all that AI.
With voice interfaces and voice assistants taking center stage across multiple industries, the demand for DSP coders and anyone with real knowledge in application development will expand even more. Finding companies that not only deliver the solutions but also have the experience required to build hardware and software solutions for audio applications — professional and consumer — and are able to support custom development in audio processing and programming is not an easy task but is more requested than ever. In fact, that is why many brands settle for me-too designs, and are not able to create differentiated products.