While self-playing instruments have existed for centuries, there exist almost no self-playing woodwinds. We decided to change that by creating Fleet Flute, the first (that we know of) self-playing flute. While partly just a cool project, Fleet-flute also has real applications for musicians with disabilities. Fleet-flute removes most of the fine motor control from flute playing, making musical expression accessible to individuals with Parkinson’s, paralysis, or other motor impairments.
What it does
Fleet flute can automatically press the keys of a flute, while all the musician has to do is blow. A set of six servos is capable of controlling up to 9 different keys at a time, fully covering a 32-note range. The servos are connected to either one or two lever arms each by nylon tendons. As a servo rotates, the tendons transmit the servo's motion to the lever arms, which pivot to depress the desired keys. An on-board processor sequentially plays the proper fingerings of a preloaded song, while the musician blows into the instrument with the appropriate airflow.
How we built it
Fleet Flute was designed using SolidWorks CAD software, starting with an existing model of a flute found on GrabCAD. The pivot arms were 3D printed, while the rest of the frame was laser-cut out of plywood. Spring wire was used to create hinges for the lever arms, and nylon fishing line was used for the cable tendons. The software runs on an Arduino Elegoo, while Sparkfun Motor Drivers provide power to the servos.
Challenges we ran into
We initially hoped to use only micro-servos in our design to minimize space requirements. However, we were only able to aquire one micro-servo, and had to depend on full-size servos for the rest of the keys. There was not enough available space to easily fit one servo per key, so we had to double the number of keys that each servo could control. Our first design relied on a series of cams to provide independent control over each lever arm, but we quickly ran into issues with cable routing as the fishing line slipped out of its intended groove. Fortunately, we were able to find a much simpler solution by connecting multiple tendons directly to the servo horns.
Accomplishments that we're proud of
Honestly, we're just proud that the machine can even play a note, let alone 30 of them! While we took on a pretty ambitious project, especially for a two-person team, we were able to divide up the workload between our complementary areas of expertise (mechanical and electrical/software) while still sharing plenty of ideas and suggestions with each other.
What's next for Fleet Flute?
While our first prototype is effective, it still needs a considerable amount of work and polish to be market-ready. The next iteration will definitely have an easy way to adjust tendon lengths on the fly, to avoid the painful calibration process that we had to go through the first time around. Additionally, there is lots of room to expand the software capabilities, such as loading songs directly from MIDI files or playing along with accompanying musicians. Finally, the shape of the device will be improved from a bulky box to an ergonomic form-factor that rests comfortably on the musician's shoulder.