Reading about tin whistles I found a reference to Warbl, a MIDI “wind controller” instrument that emulates bagpipe chanters, tin whistles and similar fipple flutes.
Unfortunately the commercial version of this relatively simple device is only available for a toe-curling $250 plus tax from its inventor Andrew Mowry in Bend, Oregon. Fortunately he has released the schematic and kicad circuit board design under the CERN Open Hardware Licence v1.2, and the firmware is available on github.
Design is based on a 32u4 AVR board running at 8MHz at 3.3V, totally USB-powered. The firmware is a standard Arduino sketch. There are an air pressure sensor, a few infra red detectors taking the place of holes, a few control buttons, a four-contact headphone socket for something called a “bell sensor” which is only used for emulating Irish Uilleann Pipes, and a USB socket. The inventor also supplies an ingenious and very simple bellows/bag which can be used for bellows-driven instruments.
I’ve thought about it, even gone so far as to 3d print a mouthpiece for an idea I had which by the time it was done printing I’d more or less convinced myself it wouldn’t work, and have yet to test it.
If you think $250 is toe curling, some of the other MIDI wind controllers on the market will wrap your toes around themselves several times. Though there are some cheaper electronic wind instruments out there such as the $100 Carry-on (which uses recorder fingering). And there are some other DIY ideas and designs out there, for instance Build the REMI mk2 - DIY Electronic Wind Instrument (EWI MIDI Controller) and Gordophone: A Series: Basics of DIY Wind Controllers . Google will lead you to more.
One thing I think I’d dislike about the REMI is its use of touch sensors for finger “holes”. But I don’t think I’d want it responding every time I merely came into contact with a “hole”; with a real pennywhistle or recorder you have to firmly close the hole. And in fact you can get altered pitches by half holing, which would be hard to emulate with almost any electronic controller, I’d think. Maybe. Maybe you can make a touch sensor that accurately distinguishes from firm pressure over the full hole versus half holing versus just brushing up against it. I don’t know.
The reflective optical sensors used by the Warbl look interesting, nice tactile emulation of pennywhistle finger holes, but again it’s not at all clear to me how well they can emulate the behavior of acoustic holes.
As a former clarinet player I’m used to actuating mechanisms to control holes, not just covering a hole with the finger. A clarinet- or sax-like controller would pretty much have to use something like tactile switches to emulate those intruments’ multiple keys.
I suppose I should try costing a build before attacking the price label. But it does seem a high price for a pressure sensor, a cheap microcontroller, a few IR detectors and some pushbuttons. The enclosure doesn’t look too amazing and I imagine a suitable length of pipe would not break the bank. There’s also an LED but I don’t think I missed anything else out.
Andrew Mowry is a professional instrument builder, not a dabbler, so I would have a lot of confidence in his shipped products. His design looks clean and he’s been responsive to demand for a whistle mouthpiece.
But I could buy a decent synth or guitar for that price.
A distinctive characteristic of many traditional wind instruments is the ability to shape the sound by “shading” toneholes with your fingers. WARBL’s infrared optical sensors detect how far a finger is from each tonehole, and WARBL uses the sensor information to send MIDI pitch-bend messages that allow for realistic slides and vibrato. Algorithms for cancelling the affects of ambient light mean that the sensors are accurate in nearly all lighting conditions.
Which I take to mean the sensors are inaccurate in some lighting conditions.
Now, I suspect that would be bright lighting conditions, and that suggests you’re OK inside any Irish style pub I’ve ever seen. It also suggests maybe you’re not OK outdoors in sunlight. But OK, where really would you be using a MIDI pennywhistle or bagpipe? Probably in the studio. Probably not in a pub session and likely not busking under bright skies.
It sounds like they’ve put a lot of time and effort into the software, but that they’re relying on software solutions to difficult problems. I mean, they’re trying to use algorithms to make optical sensors behave like acoustic finger holes, and I’m skeptical of their having a really satisfactory result.
I think the best obtainable digital solution might be some kind of high density contact-sensitive sensor array. But I suspect that really would be expensive, even if it’s possible.
And that whistle costing $215, it’s ridiculous. But their shop is in Berkeley, hipster central.
I don’t want this to turn into a pissing competition. I concede that people with money to spare can spend as much as they like. The same website sells a tunable D whistle for a far less whince-inducing £14.95, tax included. Which is nearly three times more than I paid for a perfectly acceptable Walton’s D whistle in brass.
Yes, my £5.49 D whistle, like most whistles, can be tuned by moving the plastic mouthpiece up and down the barrel. More significant tuning alterations can be made by shortening it with a hacksaw or (patiently) with sandpaper.
Andrew Mowry says he prototyped this controller with an AdaFruit ItsyBitsy, 32u4 at 8Mhz, 3.3V. Those are available in the UK for £9.84 including tax. It’s similar to a Leonardo, with integrated USB making it a good choice for USB MIDI devices like this.
His schematic specifies a Honeywell pressure sensor which is available from Mouser UK for £14.71 (before tax, probably.)
There are eight infra red detectors which act as tone holes. The part specified is QRD1113, available from Mouser UK at £12.70 for 10.
In addition to that I see some kind of RF filter chip on the USB pins, a voltage regulator for 3.3V, a few tactile switches and an optional headphone socket for emulating the use of the chanter bell in Uilleann Pipes (it isn’t actually an audio socket, just an ad hoc choice of connectors.) Finally an 8MHz crystal for the microcontroller is specified in the schematic but it, like the voltage regulator and the USB conditioning chip, is most likely already on the Adafruit prototyping board.
In the end I was able to order everything from Mouser’s UK website, and the overall price came in just over their threshold for free shipping. Mouser’s price for the Adafruit 32u4 8MHz prototype board is £8.36, £10.03 after tax. Basically the same price as Digikey but I save on shipping by bundling it into the Mouser order. Total cost with shipping, £42.92, tax included.
So I really just want to see if I can build a usable prototype wind controller based on the pennywhistle using these components. In my favour I have the open source hardware and software instructions donated to the world by the extraordinarily generous and talented Andrew Mowry, inventor of the Warbl. On the other hand, unlike Mister Mowry, I’m incorrigibly lazy and not that bright.
I didi it for a few bucks or Euros with an ESP32, some polsternails and some pipes. Beside the ESP32, I used the wrong Breath-Detector. There are some on the market which act as a differential detector. Very scary… the get the values out of it, I had to use an Analog to digital converter which is able to measure “differential” valtages… ADS1115 did it for me. I believe, my 3D-Models could help out to start quicker. MiniWI/MINI-WI-ESP32 at master · ErichHeinemann/MiniWI · GitHub … sorry the video is in german …
I have used a Airpressure-Sensor which ends which “DP” in the name, don´t take them. use a model with “GP” in the Name. For the Touchsensors I have used an MPR121-Module… You could use my code as a base for Your own “making”.
One thing I wonder about diy breath controllers is, just how efficient could it be to make the cheapest controller possible, one that’s just a glorified MIDI expression pedal and nothing more? Something for use in music production, just sending MIDI CC or CV. Something that’s not meant at all to satisfy an experienced wind instrument player, but just the bare minimum to perform or record a more expressive performance.
For this specific use, you could even set it up as a player piano of sorts, with a single button to move to the next note, to have an expressive performance across that dimension. I kinda want to build one of those for myself haha.
The basic principle is a sealed vessel containing the pressure sensor, into which a balloon intrudes. A tube from the mouth introduces your breath into the balloon. The harder you blow the more exhaust air enters the balloon, temporarily raising the air pressure in the sealed vessel. There should be provision for draining saliva and condensed moisture from the balloon (gravity will do). This sealed design keeps the sensor from being soiled and clogged up, as well as keeping atmospheric water vapour pressure from affecting the readings.
Here’s a guy with a similar idea.
Possible variants include exploitation of Bernoulli’s Principle (Wikipedia link) and using a “suck-blow” system rather than measuring pressure in only one direction.
The production version of the Warbl (invented and sold by Andrew Mowry of Bend, Oregon) has its own long slim PCB, but I’ll be using the ItsyBitsy 8MHz 3.3V prototyping board so my version won’t fit into a slim whistle-like housing like the production Warbl. The ItsyBitsy is 0.7 inch wide (it’s American, so inches.) This is 17.78mm so I’m going for a tube with 20mm internal diameter and hoping it will fit. I’ve ordered two tubes: one is aluminium for £9.99 and the other is transparent PVC for £6.99, both prices including tax. I’ll see which is easiest to work and which is nicer to play. The length of the PVC tube is 50cm and the length of the aluminium one is 30cm. 30cm is a good length for a whistle.
I still don’t have information about the exact build of the coupling used by the open source hardware Warbl design for the air pressure sensor. This is probably due in part to my ignorance of kicad, and in part to my preference for doing absolutely everything on my phone. There are online kicad viewers but I really should install kicad on my laptop and learn how to use it. The key point is clearly to use an impermeable membrane (a balloon) to keep the pressure sensor sealed away from the messy realities of human respiration while mechanically transmitting the pressure through the membrane, but I’d probably learn a lot from the details of how Andrew Mowry decided to do it.
My ability to follow spoken German is poor, but there are English captions which (though not idiomatic English) are easy to understand. I’ll need to take a closer look at how you isolate the pressure sensor from spit and whatnot. Thanks for posting this, it’s really encouraging.
