Voices from the Ether


12 October 2016

Even before we are born, our mother’s voice is the fundamental medium through which we receive information about the world around us. It slows our heart rate at seven months in the womb, regulates our sucking rate at one day old, and exerts an influence on brain regions dealing with emotions, rewards and self-awareness into adolescence.

Given these human instincts, it’s unsurprising that the voices of consumer-oriented artificial intelligence skew female, among them Siri (Apple), Cortana (Microsoft), Alexa (Amazon), Hidi (HTC), Amy (X.AI), S Voice (Samsung) and Google Now. Notably, these AI personalities are designed for one-on-one relationships with end users that imply patient listening, considered responses and emotional intelligence. Electronic devices have been talking out loud to us since the 1970s, but no one seems to have fallen in love with Speak & Spell; it took the voice of Scarlett Johansson in the 2013 film Her to make Samantha a plausible non-object of desire.

Not only are women’s voices overrepresented in AI, but they also predominate as ambient sources of guidance on public transport (Emma Clarke on the London Underground; Emma Hignett on London buses; Carolyn Hopkins on the New York Subway), in yoga classes and among the ASMR community, whose members watch lengthy YouTube videos featuring whispers or murmurs recorded on 3D binaural microphones to stimulate euphoric tingling in listeners’ skin. In these contexts, we hear the incorporeal female voice as an expression of personal attention, even when it repeats the same pre-programmed message to anyone within earshot. If our bus is diverted to a new route, we assume that Emma will sort it out for us.

As a design tool, the female voice can make an AI system appear more intelligent than its programming would indicate. Yet when women speak in scenarios that require real intelligence, they attract far more criticism than men. In the world of podcasting, women are regularly upbraided for vocal fry, upspeak and other speech characteristics perceived as immature or unprofessional. Contemporary user-listeners seem more comfortable with the female voice as a disembodied entity than as one connected to a correspondingly female brain – perhaps we need the designers themselves to pioneer a new-WAV feminism.