Conference Agenda

In 2013, Bob Sturm conducted a large-scale survey on trends in Music Information Retrieval (MIR), a burgeoning field since the late 1990s. Among his findings was the outsized presence of one particular dataset: GTZAN. Created by George Tzanetakis in 2002 for the purposes of automatic genre classification, this dataset contained the extracted statistical audio information of 1000 songs from ten genres. Other researchers found this dataset’s size and scope just right: small enough to be manageable, big enough to have some variety. He shared his dataset with anyone who asked for it, and it became among the most-used datasets in MIR in the 2000s. Despite its ubiquity, few researchers listened to the dataset; most used its data uncritically to train and test their systems. Unknowingly, these researchers disseminated the dataset’s foundational premises, or “ground truths.” Critical data scholars (Kang 2023; Mau 2019; Jaton 2021) have recently highlighted the importance of such ground truths, which constrain the questions that can be asked of data, as well as what can be done with it. The main problem with GTZAN was that researchers worked on the assumption that its statistical markers corresponded to audible markers of musical genre. But, as this paper shows, it was never clear whether these extracted features were commensurate with audible features of the music — features that scholars of genre have typically viewed as socially contingent (Gelbart 2022; Holt 2019).

This paper advances two claims. First, GTZAN’s outsized status in early ML research demonstrates in material terms how ideological assumptions are built into our present-day digital knowledge. Recommendation systems learned their ground truths from a computationalized sample of almost exclusively Euro-American culture, reinforced by algorithmic logic that recreated familiar patterns of cultural dominance. Second, this story serves as a warning that scholars should pay close attention to the history of our digital musical culture. It was made by a number of actors with particular agendas and biases, some of which persist in our current infrastructure. How do these infrastructures transform or occlude social relations mediated by music? And how might new principles of circulation invisibly reinforce older aesthetic and political hierarchies?

In 1877, Dwight’s Journal of Music reported on “the telephone revolution in music:” One day, houses in the United States would be supplied with music by telephone, like with gas or water. This vision stems from early experiments with the telephone. Beginning in the States in 1877, real-time transmissions of concerts and opera performances by telephone demonstrated the wondrous potential of the new device—even before telephones were largely used for verbal communication. These transmissions turned concerts and opera houses into spaces of scientific experimentation, which contributed to establishing the technology in society. They enabled listeners to experience live music from afar and pursued a democratic approach. Initiated by technicians and often presented as a strange episode in telephone history, these transmissions have long been overlooked by musicology. However, the phenomenon was an essential step in the history of listening and broadcasting, anticipating later phenomena ranging from radio to today’s music livestreaming.

The paper considers different sites in transnational comparisons. Bringing together reports on music transmissions in Berlin, Budapest, and the United States in the late nineteenth century, the talk evaluates the role of this early broadcasting practice by analyzing it through the lens of listening history and infrastructural concepts. With this twofold perspective, the paper points to the ambiguous effect the transmissions had on its listeners. First, I conceptualize this use of the telephone network as an “audible infrastructure” (Devine/Boudreault-Fournier 2021) to explore its connective qualities regarding institutions, musicians, and listeners going beyond distribution. The transmissions can be considered a unifying force bridging distances with music and a supply network for music. Second, building on Melissa van Drie’s work on the Parisian théâtrophone (2015/16), this paper takes the listening experience into account and questions whether this audible infrastructure could reach audiences on the perceptive level. By embedding the phenomenon in the discourse of listening history (Thorau/Ziemer 2019), I highlight the isolating and disconnecting effect of the listening mode. I argue that the claimed connective effect of the transmissions did not meet the listeners’ perception. These music transmissions via telephone rather appealed to skilled listeners and missed the democratic approach.

In 1895, the Philadelphia Inquirer published a sketch of a man and woman in a horse-drawn carriage; in it, the woman drives the vehicle with an assertive posture, her windblown hair suggesting a sporty pace and her male passenger glancing at her with surprise and intrigue. This image encapsulates the shift in perspective as women emerged as a new class of vehicular drivers at the turn of the century. Around the same time, carriage makers Studebaker Brothers and Austin, Tomlinson & Webster developed a novel marketing strategy—they routinely mailed their clients parlor-style sheet music with vehicular themes, such as “The Wagon Carol” (1884) and the “Studebaker Grand March” (1899). These songs were product advertisements that also functioned as domestic entertainment. Studebaker invested deeply in their musical campaign, commissioning songs with targeted messaging. Some songs, like “Wait for the Wagon” (1884), suggested that customers could realize social aspirations through vehicular purchases. Other songs, like “Richard and his Sweetheart Nell” (1892), suggested that carriage ownership was a visible demonstration of financial security and marital readiness.

Studebaker’s marketing strategy is more broadly connected to trends in American entrepreneurship at the turn of the century. I argue that carriage makers borrowed this promotional technique directly from the political arena where presidential candidates used campaign music to rally support (Gorzelany-Mostak, 2018). I also trace the adoption of Studebaker’s merchandising tactics by other businesses, such as Garland Stove Manufactures, Schilling Corset Company, and the Emerson Drug Company, who similarly circulated sheet music that equated their products with an aspirational lifestyle. This study contributes to growing research on the role of music in marketing and advertising (Deaville, Rodman, Tan, 2021), showing how music has been used to shape the emotional responses and perceptions of consumers. Focusing on the less-studied, early development of musical marketing, I contend that Studebaker’s promotional songs offered women a chance to envision new roles as they straddled the parlor and reins. More broadly, this project shows how nineteenth-century, American entrepreneurs were experimenting with strategies to leverage music for monetary and cultural profit.