Yes, this looks like an ordinary stopmbox, but it is reprogrammable.  Can I put this massive "prototype" disclaimer over any photos of me tagged on Facebook? No? Photo courtesy the OWL folks.

Yes, this looks like an ordinary stompbox, but it is reprogrammable. Can I put this massive “prototype” disclaimer over any photos of me tagged on Facebook? No? Photo courtesy the OWL folks.

There are stompboxes. They are — for lack of a better word — foot worthy. You can step on them, in a way that is less possible with a computer. (Well, sure, somewhere amidst an endless spinning color pinwheel you may have wanted to step on your MacBook Air, but then thought better of it – financial investment and whatnot.)

Then, there are computers. They can do everything. That stompbox is one particular distortion effect. And it is always just that one distortion.

But what if you could have both?

As embedded technology continues its march toward greater user friendliness, lower cost, and greater sonic powers, it seems the time is right for hardware that combines the durability of dedicated sound gear with the open-ended potential of computers. That is, it’s not really clear where the computer ends and the stompbox begins.

OWL isn’t the first project to take on this dream, but it’s looking more practical than those that came before.

The project promises open source hardware, with open code, that can be reprogrammed into new sound effects simply by uploading new code. As with a new generation of low-power tablets and phones and the like, there’s an ARM chip at its heart. (The ARM Cortex M4, to be exact.)

If you’re a guitarist who writes your own C++ code – yes, there’s actually a sizable group of those – you can have a ball making your own DSP routines. If you’re not, OWL promises a library of patches, presumably growing with more contributions from the open source community.

There’s not a whole lot to look at at this point – while they’ve got a GitHub repository going, it includes only a little bit of sample code. But in the video, the results look impressive, perhaps enough – given an experience team – for some to go ahead and take the leap of supporting the crowd-funded Kickstarter project.

Patches load directly via USB – so reprogramming the pedal is a pretty easy affair for the average user. If you are a coder, you can use simple C++ without the usual mucking about with hardware-specific code. (That’s where, to me, the advantages of newer ARM chips comes in: there’s enough horsepower here that you don’t have to fret over every spare cycle, coding close to the iron. But if you do want to use specific ARM functions, those are supported in the framework, too.)

What you get in the product appears to be a no-nonsense hardware platform with the requisite jack connections and stomp-able switches, and a straightforward code framework. It’s not quite as idiot-proof as something like Arduino, but to a growing army of DSP students around the world, it’s a beautiful blank canvas.

It’ll be fun to watch this evolve. And there appears to be at least enough crowd funding to get it rolling – with additional funding “unlocking” additional work from the team on other features. See more at the Kickstarter site:

OWL Programmable Effects Pedal [Kickstarter]

The team has as its members a number of friends of CDM, including the maker of the Blipbox, and others revolving around the lovely London Music Hackspace.

I like in a way that the product isn’t too ambitious: it’s simple, uses a smart platform as its basis, and focuses on things people need.

It seems there’s more to do in this space. Years ago, the talented originator of Winamp and Reaper made the JesuSonic, dedicated hardware for effects cheekily hidden in a massive crucifix. But now, that sort of technology can easily hit the mainstream – with or without weird religious iconographic housings. The other logical direction seems to be more traditional computers running Linux, the sort which could take uploads dynamically using tools like Pure Data, without having to reprogram the pedal between each set. But both directions – embedded computers and dedicated hardware – hold potential, and both could be reprogrammable. OWL could be the herald of things to come, and if successful, the first real case study in making those things work.




  • mdur

    Looks cool. Reminds me of the open music labs MicroDec in pedal form.

  • Shebö

    OWL indeed isn’t the first project to take on this dream, however I think stuff around the Spin FV-1 might still be more accessible. I’m not sure, but I’d say the FV-1 is easier to program, and a FV-1 development board requires only small modifications and is currently available for around 80 EUR. Or if you’re good at DIY electronics, build a Tonepad Fraverb and add (or don’t, since the FV-1 comes with usable internal FX) EEPROM programming capabilities.

    • gwenhwyfaer

      Depends what you mean by accessible. In terms of programming, C++ is going to be a lot more approachable than FV-1 assembler language; in terms of hardware, the FV-1’s architecture prevents it from doing convolution-based effects – or, in fact, anything that takes more than 128 instruction cycles per sample (in an instruction set that’s optimised for particular kinds of effects, and which converts conditional jumps into skips over each jumped instruction to keep the 128-cycle time constant), so while there might be less to master there are also some pretty immovable limits. Moreover, you can work out the basics of your algorithm on any old computer for which you have a C++ compiler, and just change your audio interface code (and optimise if necessary) when your OWL box arrives – in fact, if I were the OWL folks, I’d be looking at releasing an API which can interface equally well down with OWL or JACK – but I’m not, so never mind :)

    • Sean Costello

      I would be interested to see if the Cortex M4 can outperform the FV-1 in the sorts of applications the FV-1 is suited for (i.e. “allpass loop” reverbs, overlap/add pitch shifting). For algorithmic reverb applications, memory access speed is key, and if the Cortex M4 takes a few cycles per memory read, this could end up slowing things down considerably.

    • Digital Larry

      I’m developing a program that promises to make creating patches for the FV-1 quite a bit simpler. It doesn’t eliminate any of the FV-1’s inherent limitations but it speeds up patch development by probably 100 times.

  • peter venus

    minus the DIY-hardware, it reminds me on the MOD, a pedal-board that hosts (free) LV2 plugins

  • Ross Healy

    Snazzy FX Ardcore?

  • Jesse Engel

    Hey Peter, you mention the linux approach at the end of the article, and for all those interested, just thought you might want link

    Satellite CCRMA (linux/pd/rasberry pi or beagle board)

    I’m not sure if anyone remembers the Line6 DSP ToneCore pedals, but C++ is definitely easier than assembly, so this entry seems valuable :).

  • Taylor

    This looks sweet! It would be fun to port my app’s audio engine to the Owl.

  • just passing

    Worth mentioning that the Cortex M4 has DSP extensions in quite the same way as, say, a dsPIC has DSP extensions. The dsPIC’s MAC instructions, in full flight, can do a 40+17×17-bit multiply-accumulate, two loads, and a round and store in a single instruction / cycle; it also has hardware support for zero-overhead loops, circular buffers and bit-reversed addressing, the latter a big win for FFTs. Moreover, even when not using the DSP pipe, every dsPIC instruction is memory-to-memory (dsPIC memory is basically 32,768 registers), so the dsPIC can do with a single instruction (and in a single cycle) what it would take an ARM two or three instructions, along with interlock considerations.

    On the other hand, the Cortex M4 is 32-bit, can write code to and run code from RAM, and has floating-point (with 32 separate registers!) and SIMD instructions (including SIMD MACs that accumulate two 16×16-bit products into a single 64-bit accumulator). So there are goodies on offer here too. (I dearly wish that Microchip had extended the dsPIC architecture to 32 bits, or even 24, instead of buying in a MIPS core.)

    • just passing

      …er, that should read “the Cortex M4 *doesn’t have* DSP extensions in quite the same way”…

  • AutoStatic

    The Linux Approach 😉

    Guitarix on ARM:
    Using a Raspberry Pi as a virtual guitar amp:
    MOD (which uses Arch Linux on an Intel Atom board):

  • A

    I was just looking at the Zoom MS-100bt, a reconfiguarabe (via bluetooth) multi-effects and while it’s not open source, I wonder if they would open it up to external development? They even have an AppStore. It seems like a winner compared to the similar digitech offering. Absolutely charming pixel graphics display. 😉

    Without a screen, reconfiguration appear useless to me? I do get the appeal of an open platform to allow software creators easier acess to make pedals, or other hardware tactile interfaces, but…

    The cost seems to be in designing and implementing a nice interface, not the computing power. Personally, I’d rather put up with using a computer as the brain to an exceptionally well designed controller, if it kept the costs down.

  • Seb C

    hi guys, cool project are you using a STM32F4 chip (using integrated 12b ADC) ?