Is it true that there are only 11 real audio designers in the world? Are today's low-cost, high-performance general-purpose microcontrollers pretenders to the DSP throne?
The headline to this column seems a little crazy, doesn't it? I thought so as well until I met up with Paul Beckmann of DSP Concepts. I was talking to him about a startup he helped called SoundFocus that boasts a smartphone case with amazing audio you can tune to your hearing and the environment you are in.
Full disclosure, I'm consulting for ARM and looking for cool new products with ARM processors in them, so there was a method to my madness! The Amp from SoundFocus is a really cool product. It has amazing audio crammed into a very small space, which formed a key pivot point in my conversation with Paul.
The following graphic provides a visual explanation of why Paul estimates there are only 11 real audio designers in the world:
Audio engineers tend to originate from one of three disciplines -- sound engineering, embedded software development, and DSP engineering. As we see in the Venn diagram that Paul constructed, the end result is the magic number of 11.
I've seen this kind of data before in a survey undertaken by EETimes. That survey showed that engineers have to work in as many as five disciplines at any point in time, yet they only study one in college.
So, if we accept that it's incredibly hard to find people skilled in multiple disciplines, what are design teams to do? Paul's solution was to build a software tool called Audio Weaver for very fast audio prototyping and design. This tool is based on MATLAB and is based on drag-and-drop components. In just a few hours, it's possible to design a system and prototype it on a board, and this is exactly what the two young founders of SoundFocus did. Here is what Audio Weaver looks like with an STM32 Discovery target board:
It really is astonishingly easy to use Audio Weaver -- which you can obtain from DSP Concepts -- and it's free to use (with a royalty for production).
Paul's other heretical question shown in the title to this column was whether you really need a dedicated DSP to do good-quality audio design. The simple answer is "Probably not." Paul grew up on DSPs at MIT where he got his PhD, and he went on to work at Bose, so he knows audio. The fact that Paul is so conversant with this field underscores the shift that may be happening in the audio market.
Paul told me that 80% of his consulting work is no longer focused on the use of DSPs, but on ARM-based devices like the STM32F407. This is really good news for designers, because now you can design audio products with lower cost parts that consume less power, all with tools that allow you to prototype in a day.
Now, please don't think I am saying that the DSP is dead, because Paul's own benchmarking -- as described in this paper that he gave at AES this year -- shows that in some applications the dedicated DSP is still the King (Analog Devices' SHARC in this case). In many application areas, however, today's low-cost, high-performance general-purpose microcontrollers are pretenders to the DSP throne.
I have a feeling that this is a tipping point in getting better quality audio in cheaper products, and that has to be a good thing. What's your take on this?