[Interview] Stefano D’Angelo PhD, Italy (Digital Signal Processing Engineer, contracting engineer)

Stefano D’Angelo, PhD, is a DSP engineer from Italy, he has created many popular audio synthesis VST plugins, and has authored many scientific papers in the field of audio DSP and virtual analog modeling (making digital emulators of the analog designs). Today we will talk with him about digital vs analog processing, dynamic range limits in audio processing, commercial and technical aspects of VST development, and navigating between scientific career and contract work such as commercial VST development and analog design contracting.

 

Did you ever work in academia full time? It seems you are more aligned with the industry, but however you have 200+ citations in scientific literature. What is your research/development ratio in general?

Well, if you consider doctoral studies as work then I did at the time, otherwise I didn’t.

In general, my work is often at the edge of what is currently known, so it is rather natural for me to cross the borders of the state of the art, find out some original solution, and then integrate my findings into real-world products. When that happens, 90% of the time I try to publish results.

It might sound odd, but this means that I tend to actually do research when clients’ requests present new challenges.

This is an elaborate way to say research and development go hand in hand and I can’t easily distinguish between the two.

 

So you earn solely by your contract work? Any plans to get a full time job at university?

I also founded and run a company called Orastron (www.orastron.com) whose aim is to create algorithms and software tools to help companies and independent developers build and/or improve their audio products. We are still in somewhat early stages, though.

At the moment I don’t plan to do anything else, this is already quite a lot.

 

You have done a lot of remodeling analog circuits into a digital domain. What is your overall feeling on this? Lot of audiophiles prefer analog, but analog really hasn’t made much progress it seems. What is your take on this?

It is a fact that analog is much more limited by physical constraints than digital, and also that digital technology has progressed tremendously in the last couple of decades. On one hand digital audio, and music technology in particular, are relatively small markets that are often overlooked or neglected by big tech companies, which makes it harder to develop good products. But on the other hand it’s plain ridiculous to think that analog can be the way forward in 2023.

BTW, in case you wonder, I can peacefully state that well-engineered digital products can and do sound better than analog equivalents. A lot of affection for analog products is due to nostalgia and “magical thinking” today.

 

Do you think the digital domain will increase its resolution and bit depth to enable more advanced DSP algorithms? In comparison, video processing and GPU power has increased much more in the last decade.

Bit depth is not a limiting factor, 20 bits and 40 kHz sample rate are enough to cover the whole human hearing range. We could perhaps use more computational power and faster memory for very specific applications like spatial sound, physical modeling and AI/machine learning, but honestly I believe that it would be important to also concentrate on the human factors involved (read: knowledge, competence, etc.) both among users and product makers.

 

You have also created many VST plugins – in which language and IDE do you do those and are there any difficulties following the VST SDK by Steinberg?

Honestly I only take care of the purely DSP part, so the languages and IDEs are not overwhelmingly relevant for my work (well, within certain boundaries). But in general, everybody still uses C/C++ and sometimes machine-specific assembly, Visual Studio and Xcode are very common, and Steinberg’s VST SDK is… full of surprises, to use an euphemism.

 

I know you developed many plugins for other companies like Arturia, but do you also sell some of your VST plugin?

No, or at least not yet.

 

Some companies, Waves for example, have totally moved to the subscription model. Do you think this is the future?

I don’t really have an opinion on this. Maybe.

 

Do you think LADSPA on linux is a good alternative to VST? There don’t seem to be as many plugins available, why is that?

LADSPA is effectively dead. LV2 was meant to compete with VST, and while I believe it’s technically superior in many ways it didn’t catch up for a number of “psychological”, social and marketing factors (it uses semantic web technology which most developers don’t understand or like, it’s not really backed by any company, etc.). CLAP is the new kid on the block backed by a few important companies (u-he, Bitwig, I don’t know who else) and, while I’m still skeptical about it – haven’t really played with it yet, it would be probably a good thing if it succeeded.

 

When designing “realtime” audio tools to run on desktop computers, do you rely on ASIO architecture or something else? Can ALSA be compared to ASIO in terms of performance like latency and CPU utilization?

So far I have avoided anything platform-specific. Anyway, ALSA is fully capable for pro audio and Pipewire is amazing.

 

In general, what you think will be the most interesting IDE and language for audio development for desktop computers  in the near future?

My company is developing one called Ciaramella (ciaramella.dev/). Keep an eye on that. 😉

 

And what about audio applications for embedded design? What platform do you think is promising? Raspberry Pi, Android and iOS smartphones, or some custom low level ARM based hardware? 

All of them and I’m sure there will be more coming. Keep an eye on the Daisy Seed (www.electro-smith.com/daisy/daisy) – it’s plain great for developing audio products. Mobile OSes are already in the hands of consumers, with iOS having great audio performance and Android lacking in the low-latency department. Also I think that the web as an audio platform might prove interesting.

Where can we find your plugins, other products, and follow the science you produce?

www.dangelo.audio/ and www.orastron.com/ .

 

Would you be willing to connect with our readers on any social media platform?

Yes, just add me or follow dangelo.audio and Orastron on Facebook, Linkedin and Youtube.