Bodies Reappear as Action

Abstract

What forms of echoic, intersubjective relations can we establish with synthetic voices. How can these affect the production of our own voices?

Through a reflection on a series of my own synthetic speech performances, this article considers how notions such as body, presence and action may be associated with the field and practice of voice synthesis and how synthetic voices, as the quintessential disembodied voices, may access the space of bodily co-presence constitutive of the performance event.

We haven't yet naturalized voice-based digital assistants, yet the companies that market them are already going to great efforts to bring them into our homes by means of always-on, always-listening devices such as Amazon Echo. Their determination reflects an attempt to make their services seamlessly disappear into the fabric of our quotidian affective interactions.

Disappear, here, would mean that when speaking to these devices, we would feel as though we’re interacting with another person as opposed to a computer. Consequently, hearing the sound of these voices would not necessarily remind us of their affiliation. As Amazon's own structure becomes increasingly complex, its voices are built to be more and more 'natural sounding'.

Rather than pursuing synthetic voices as naturalistic representations of the human voice, whose value resides in their ability to conceal the processes of their own production, this article seeks to listen to these voice's skewedness and technical genealogies in order to discern how their own material specificity and presence unfold.

Notes

1 I am using ‘cite’ as a translation of the Spanish verb citar, which has the additional meaning of making an appointment with someone. For a thorough development of this double meaning see Garbayo-Maeztu (2016).

2 For a detailed discussion on vocal timbre and race see Eidsheim (2009).

3 A particular case is that of the singer known as T-Pain, whose musical proposal and artistic persona rely entirely on the use of ‘Auto-Tune’, a vocal effect popularized by Cher’s single ‘Believe’ in 1998. For an interesting exploration of recent cases of vocal manipulation in music production see Borkowski (2014).

4 ‘Natural-sounding’ is how IBM refers to its synthetic voices in IBM Watson’s developer documents.

5 An example of this approach would be artist R. Marcos Mota’s (2014) performance ‘Al menos tres canciones de un travesti en el espacio’ (At Least Three Songs of a Transvestite in Space). The artist was carried into the stage (by, among other people, myself) wearing a black, mermaid-like dress, covering their whole body, including their face, and left on the ground. They then remained still for the whole duration of the piece: a speech synthesis based performance with musical interludes, after which they were carried out of the stage.