The oleaginous voice: Auto-Tune, linear predictive coding, and the security-petroleum complex

In the twentieth century, exploration geophysics and digital vocal processing each underwent key technical transformations based on formally similar signal processing techniques derived from prior work in statistics and information theory. Engineers at Texas Instruments (TI) and the Massachusetts Institute of Technology (MIT) developed deconvolution filtering as a way to model geological strata from seismograph recordings in order to facilitate oil exploration. Meanwhile, engineers at Bell Laboratories and Nippon Telegraph and Telephone (NTT) used similar techniques to digitally encode and transmit speech signals more efficiently - a method known as linear predictive coding (LPC). These cognate approaches to modeling geophysical and vocal signals provided a practical basis for subsequent forays by oil industry figures into the world of digital audio. Tracing the histories of four technologies - LPC secure voice transmission terminals, the TI Speak & Spell educational toy, automatic speech recognition, and the vocal pitch correction software Auto-Tune - I analyze the role of security interests in facilitating spillovers between technologies of oil and the voice. I argue that, by approaching both geophysical and vocal signals as matters of national security, the United States' defense and intelligence services furnished the bulk of socio-economic connective tissue between petroleum geophysics and vocal processing research.