Troy Ounce

One question hangs over the process of resurrecting these works first made for my PhD portfolio: is it worth bringing them all back? Troy Ounce was one of the earliest pieces, and revisiting it reminds me how those first experiments seeded ideas that later became central to the portfolio.

Financial data (alongside weather data) was one of the first things I noticed online that had a rate of change lively enough to suggest a musical potential. There were already many sonification works mapping financial values to musical parameters in inventive ways, although some fell into my least favourite category: higher value = higher pitch. Rigorous, no doubt, and potentially useful as audio graphing, but musically unsatisfying.

Troy Ounce marked a breakthrough. After auditioning several unsuccessful attempts that treated the data as direct musical parameters, I realised I could wrestle control of the compositional framework back, using the data to drive a process instead. This led to an approach I recycled across the portfolio. Drawing inspiration from Janek Schafer’s multi‑arm turntable, Leif Inge’s 9 Beet Stretch, and the “400% slower” Paulstretch trend, I designed a system in which movements in the data act as the playhead of a virtual record, a soundfile frozen by a phase‑vocoder algorithm, nudged forwards and backwards by fluctuations in the gold price. It unlocked the idea of data sonification navigating a soundworld (of my design) rather than generating melodic material. The virtual record, providing an artistic pun, is Gold! by Spandau Ballet. It may remain something of a bagatelle in the portfolio, and the sometimes harsh sound of time‑stretched 80s pop may preclude it being described, like others, as an ambient website, but I deemed it worthy of revival.

Things have moved on considerably since the original version. Yahoo Finance—the initial data source—died a death during the writing up of my thesis, and I resorted to web‑scraping financial sites for the submitted version. For this new iteration I found a handy API providing spot gold prices in several currencies every 30 seconds, all for free and straight to the browser (no scraping or cross‑origin contortions). Of course, like all these pieces, I can imagine logging on in a year’s time only to find the endpoint discontinued.

One of the gratifying aspects of revisiting these works is the dialogue with my past self: ironing out mistakes or missed opportunities. In hearing the new version of Troy Ounce—loading and re‑loading the page as each change was made—I realised that, as a generative work intended to be different on each listen, it had an Achilles heel: it always started at the same point. Despite the randomness of the gold readings, the opening sound was fixed. I hadn’t diagnosed this in the original portfolio, and it only became apparent when testing this web version. The solution was to place the starting points of the four channels relative to a fixed moment—the release of Gold! in August 1983—simulating the sample wrapping many times since then, as if the piece had been continuously running ever since.

The original version of Troy Ounce is documented in my PhD thesis.