when i began my internship at the input devices and music interaction laboratory (IDMIL) at mcgill university this summer, my professor marcelo wanderley suggested that i focus on the t-stick. the t-stick is a digital musical instrument (DMI) pioneered by joseph malloch and andrew stewart at the IDMIL. it was first concieved in 2006, and has seen a few different iterations. more info on the t-stick can be found on the IDMIL website:

i decided that i wanted to develop a practice with the t-stick, as i have with the piano, and to integrate the instrument into the musical contexts i am already a part of.

marcelo initially suggested that i begin with one of the early, wired models of the t-stick, as they were currently unused and sitting in a large cardboard box in the corner of the lab. i quickly discovered that there a few glaring reasons these early t-sticks were out of commission. firstly, clear documentation that detailed the process of using, testing, and troubleshooting these first-generation t-sticks was difficult to compile. after some digging around, and help from my peers in the lab, i was able to find a puredat a patch designed to aid in t-stick configuration. while this patch allowed me to test the wired t-sticks, it was difficult for me to parse the incoming data, and use it consistently and reliably. in addition, only one of these t-sticks appeared to be (mostly) functioning. eventually, at the suggestion of one of the researchers in the lab, i decided to use one of the newer wireless t-sticks.

using a wireless t-stick first required that i learn and understand the osc protocol. osc messages are sent from the t-stick, over a network, and are recieved by a connected computer. once i was successfully recieving osc messages, i needed an interface for organizing the incoming data. i designed a pure data patch, similar to the one i had used with the wired t-sticks, which allows me to route the osc messages as i need to.

with a working t-stick and a method for delivering and sorting the data from its sensors, it was time to focus on sound generation. on the immanuel wilkins album "the 7th hand", drummer kweku plays a cymbal/bell percussion instrument which struck me the first time i heard it:

i wanted to interact with the t-stick as a percussionist, and i was heavily inspired by this sound. i needed to find a suitable synthesis method. i've had an interest in physical modelling for some time now, however i decided against the use of a physical model for this project, as the research it would require would distract me from my main goal. i then moved to an additive method, inspired by modal synthesis. while this was interesting, i encountered a few issues.

firstly, to obtain a harmonically complex sound via additive synthesis, one requires a large amount of sinusoidal oscillators. in pure data, the osc~ and cos~ objects are implemented via wavetable lookup. the computational cost of this procedure is within reason, given the relatively small number of oscillators i decided to initially use. however, i quickly noticed that, if i were to scale my first draft, my computer would have trouble running the patch smoothly in conjuction with the t-stick data routing patch. in addition, the number of parameters in the patch would've been unwieldy to use. while i could've modified and adjusted my patch to suit both my needs and the restrictions of my computer, i figured there was probably a more fruitful method. a first test/jam with some other musicians confirmed my thoughts:

it's clear that, regardless of my lack of facility with the t-stick, the sonic flexibility of my patch in conjunction with the mapping i created was lacking.

after listening to a talk given by max mathews, john chowning, and curtis roads, i had FM synthesis on my mind. while i could've modified and adjusted my patch to suit both my needs and the restrictions of my computer, i realized that frequency modulation would be a more efficient and user-friendly way to obtain harmonically complex percussion sounds. keeping the general interface of the additive version of my patch, i put together a 2-operator FM synth, with frequency and decay controls available for both the carrier and the modulator. once i had the basic idea working, i needed to devise a suitable mapping from the t-stick controls to the patch.

i created sub-patches to derive sensor count, the number of sensors being touched, touch onset, the moment the t-stick is touched, and the time elapsed after it is touched. using these data, I was able to generate percussive sounds by touching the t-stick, and introduce timbral variation after a given time is elapsed. i mapped the decay time of both the carrier and modulator to the pressure sensor (FSR), to obtain longer sustained sounds with greater pressure. i'm currently in the process of figuring out how to use the gyroscope to detune the oscillators, however i've been struggling scale/map the data from the gyroscope in a playable way.

overall, one of the biggest difficulties of this whole process has been developing the interaction-to-sound feedback loop that is so necessary for musical practice. in my experience practicing the piano, paying close attention to how the subtle changes in my technique affect sound production has played a big role in my development on the instrument. as i change and elaborate on the my t-stick synth patch, my technique has to adapt. i'm hoping that as the patch reaches a stable state, more room will be available for this necessary process.