Journal of Interdisciplinary Voice Studies - Posthuman Voices: Channels across Time and Shared Memories, Aug 2022

This article deals with the groundbreaking phenomenon of AI voice, highlighting two possible meanings that are often not problematized: the voice embedded into AI-based devices and the voice created using AI algorithms. In order to clarify the distinctions and the intersections of these two meanings, the article uses an approach inspired by media archaeology and social constructionism. It argues that AI voice as a social phenomenon is constructed by the interaction of a discursive level of representations and a non-discursive level of material practices and operations. The interaction of these two levels results in a tension between anthropocentrism and posthumanism, which is a characteristic of AI voice. Such tension is investigated through two case studies: the commercial of the smart speaker Amazon Alexa and the phenomenon of ‘voice cloning’. While the first is an example of how at a discursive level the ‘voice in the machine’ is represented as a way to ‘personify’ AI technology, the second, which consists in the possibility of reproducing the features of an embodied and personal voice, is an example of how the materialization of that cultural idea depends on the technical possibilities and material practices required by data-driven algorithms.

There is something paradoxical about the fact that while whales and dolphins produce some of the most complex vocalizations on Earth, they have little political representation or ‘voice’ and despite the success of past anti-whaling campaigns, continue to face existential threats from entanglement, ship strikes and underwater noise pollution. In this article, I argue that this paradox is sustained by a belief in human exceptionalism – exemplified by the claim that music is unique to humans – and review biological and musicological evidence that contradicts this claim. Overcoming the paradox may require more than logical argument, however, and I survey the use of humpback whale song field recordings in works of human music, analysing them along the dimensions of ‘distance’ and ‘difference’. I argue that although it is important to recognize the continuity between human music and humpback song, a more effective use of whale song recordings also requires attention to be paid to the differences between human and whale vocalizations to avoid the risk of collapsing into naïve anthropomorphism. Such an animalcentric compositional voice would operate according to the ideals of ‘difference without distance’ and ‘proximity without indifference’ to facilitate empathic relationships between humans and other animals.

In this Voicing, Ute Wassermann describes how sympoietic vocal practice brings her into resonance with the world in different ways, creating a complex network of relationships within her body between various vocal identities. Stories are told about how her many voices and the environment exist in a mutually stimulating feedback relationship. She gives examples of how her sympoietic voice collaborates with the polyphonies of other-than-human voices. She communicates with voices sounding from objects, and at the same time is influenced by them. Does her voice remain human, or will it become the other?

A voice-based autobiographical performance work, developed by deep listening to a synthetic clone of my voice, serves as a point of departure to consider the aesthetic and sociopolitical implications of the modes in which voices are represented in speech synthesis. In the context of the rapidly increasing use of voice-based devices and virtual assistants, such as Amazon Echo and Alexa, I discuss how recently developed artificial intelligence techniques are being used to produce hyper-naturalistic voices that conceal the sonic markers left by their production processes. These attempts at technically reproducing voices that are naturalized, by becoming indistinguishable from their originals, are historically situated through the examination of a series of cases, such as Thomas Alva Edison’s Tone Tests. In addition, these show how, by closely listening to synthetic voices’ technically mediated difference, we may establish forms of affective relationship with them that can favour a lasting transformation of how voices are thought and produced.

This Voicing explores the theoretical and material connections between Artificial Intelligence (AI) and voice. The Voicing is in three-parts encompassing: a theoretical introduction with a taxonomy for voice and AI, an extract from artist K. Allado-McDowell’s new work Air Age Blueprint and an interview based on such text. Allado-McDowell pioneered the field of human–artificial intelligence interaction and literature. With their book Pharmako-AI, written in collaboration with GPT-3, Allado-McDowell stretched the limits of language creation. Francesco Bentivegna worked as artist in the liminal space of cyborgean voices and recently completed their Ph.D. on voice, AI and synthetic personas. As an exercise in philosophy of AI, voice studies and artistic research, Allado-McDowell and Bentivegna move in conversation through biases, vibes and language, exploring narrative, science and theory.

Review of: The Computer’s Voice: From Star Trek to Siri, Liz W. Faber (2020)

Minneapolis, MI: University of Minnesota Press, 256 pp.,

ISBN 978-1-51790-976-5, p/bk, $27.00