Spoken For

The new “genderless virtual assistant” is not a neutral voice

“I’m created for a future where we are no longer defined by gender, but rather how we define ourselves,” says Q, a new “genderless” digital voice — a proposed addition to digital assistants like Alexa and Siri, who currently only come in male and female flavors. Q was developed in a collaboration between Copenhagen Pride, a handful of Danish technology companies, and an ad agency whose website touts its “DIY punk roots” and “empathy.” According to their Instagram profile, these creators hope to pitch Q to companies like Apple, Amazon, Google, and Microsoft.

Q is purported to speak without gender: Their voice reverberates at 145 to 175 hertz, a frequency range determined by scientists to register in the human ear as “neutral,” between the frequencies associated with cis men and cis women. “Naming a home assistant Alexa, which sounds female, can be problematic for some people, because it reinforces this stereotype that females assist and support people in tasks,” Project Q collaborator Julie Carpenter told Wired. If Q has no gender, the logic goes, their voice and its docility cannot reinscribe an existing gender stereotype.

Q, being polite and servile, sounds little like the trans human beings on which their voice is based

Q’s marketing addresses two distinct problems: first, the cisnormativity of tech interfaces, which generally only acknowledge male and female genders; Q aims to invigorate the voice assistant market with gender diversity, reflecting the existence of non-binary consumers who might hear themselves echoed. Second, the conflation of femininity with docility, which is why Alexa and Siri more often than not sound female. “Together we can ensure that technology recognizes us all,” Q says.

Q sounds placid, as digital assistants do. When I envision a body behind the voice, I see a gay flight attendant, a cis man in a traditionally feminine job making use of female speech patterns to reassure potentially anxious customers. Of course, a gay flight attendant has plenty of gender: ways of dressing and moving and speaking that facilitate interaction between the individual and the world. It is difficult to imagine a body, or a voice, without any.

Q is meant to “end gender bias” by disrupting the association between female voices and servitude. But its solution to both, in effect, is to insert a “neutral” voice into a position of total passivity. By their very lack of agency, Q occupies the same gendered position as other robots in popular culture: servile, responsive, not quite female — but certainly not male.

In the essay “The Gender of Sound,” Anne Carson locates a rubric of gendered vocal identities in Greek antiquity. These can be arranged along a binary, if not the one used today: Men, and everyone else. “High vocal pitch goes together with talkativeness to characterize a person who is deviant from or deficient in the masculine ideal of self-control,” Carson writes. “Women, catamites, eunuchs and androgynes fall into this category. Their sounds are bad to hear and make men uncomfortable.”

A robot’s voice stutters and breaks. It does not respond to a human interlocutor’s contextual changes in pitch and tempo, and speaks without much nuance or spontaneity. Because of these subtle but immediately recognizable cues, the robot’s voice registers as Other even when it vibrates at a pitch associated with cis men; it hits the ear uncannily, and identifies itself as coming from a badly gendered entity — one not in control, not at the center of the world.

Robots in the popular imagination have rarely gone without gender markers. If they have feminine bodies and voices, they tend to be either servile or evil; if they use he/him pronouns, they tend to be rendered broadly effeminate, or in possession of an outrageous and clearly overcompensating masculinity.

Q is not a neutral voice, just a neutralized one. Gender is not a texture that can be filed away when inconvenient

The first iconic robot in the movies was a replica of a woman, and was also a sex worker. In Fritz Lang’s 1927 silent opus Metropolis, an evil scientist creates a mechanical copy of a saintly woman named, rather obviously, Maria, who first appears in the film haloed by abundant light while chaperoning impoverished children. By contrast, the “fake” Maria has no virtue, no scruples. In a famous scene, the android rises from a steaming coffin to dance before a room of men, who are immediately bewitched. Lang isolates the men’s eyes from the rest of their faces and stitches them together in the same frame, filling the screen with a helplessly horny Argus. “For her, all seven deadly sins!” proclaims one audience member. Bad Maria is like a woman, only more so, and as such she leads men to ruin.

In movies, the robot is often feminized, even if she is not rendered as a woman. Star Wars‘ C-3PO was modeled after Maria’s mechanical skeleton, the “true” form to which she returns when she is burned at the stake. Played by British actor Anthony Daniels in the film franchise, C-3PO is fussy, loquacious, and definitively effeminate. His subplots with the handy, no-nonsense robot R2-D2 comprise the best known fag/dyke buddy comedy in film.

Girlish robots proliferate throughout film history: as fuckbots in Blade Runner, Ex Machina, and Westworld; as disembodied, devoted voice assistants in Her. These fembots begin life in serene docility, but mutate into Eve-like betrayers, thwarting the fantasies of their male creators when they dare to pursue freedom from Eden.

Masc bots sin differently. When grafted onto robots, alpha masculinity becomes distended and uncanny; Robocop and the Terminator supplant organic masculinity with a hilariously overwrought form of butch. Others present a beta masculinity ripe with pathos: the tragically named Alpha from Power Rangers, whose neuroses constantly short-circuit him, or Star Trek: The Next Generation‘s Data. In one episode of TNG, the humans aboard the USS Enterprise fall ill with a disease that mimics the effects of drunkenness. The security chief, a soft butch blonde, asks Data to fuck, and he does. Once recovered, she asks the android to never again to mention the encounter, as if it were a lapse in normative sexuality, a drunken bicurious mistake. “Let’s pretend this never happened” is a familiar enough line to the confusingly gendered, those who, like the robot, attract and repel the eye in the same look.

Like the badly gendered inhabitants of ancient Greece, robots are designed to speak when spoken to. They annoy when they overstep their bounds. (In the original Star Wars trilogy, a running gag is that C-3PO talks too much and must be shushed by Harrison Ford’s Han Solo; Data similarly fails to read social cues and speaks too long in too much detail.) The robots that have become commonplace in homes are also expected to remain silent unless asked otherwise. Amazon’s voice assistant Alexa lives in a body that is literally named Echo. She is not supposed to vocalize unprompted, and when she does, she terrifies her masters. In 2018, Echo owners reported a spate of “creepy” laughter emanating from the machine. “Lying in bed about to fall asleep when Alexa on my Amazon Echo Dot lets out a very loud and creepy laugh,” went one indicative tweet. “There’s a good chance I get murdered tonight.” A robot vocalizing on her own implies a threat to her master. The sound is bad to hear and makes men uncomfortable.

Q was created with the help of trans and non-binary people, who recorded their speaking voices to be analyzed algorithmically. In the project’s demo video, these people are shown in shadow, with only parts of their faces illuminated. There’s a brief moment where you can hear a chorus of trans people speaking in unison, then suddenly their voices get ironed out to the average of their tones. People of marginalized genders speak, and science sands down their particularities to give rise to “the world’s first genderless voice” — a gentle, unobtrusive presence that invites users to consider trans people, rather than cis women, their personal assistants for a change.

To transition as a non-binary person is not to settle on a comfortable shade of grey somewhere between two available poles

Q’s introductory video visualizes their supposedly neutral voice as lying between two poles: the low voices of cis men, and the high voices of cis women. On the website genderlessvoice.com, the user can drag a pulsating orb around the screen to hear Q speak in all three registers. But Q, being polite and servile, sounds little like the trans human beings on which their voice is based. Both transfeminine and transmasculine people speak in the demo video, and their voices ring in about the same range. These are not naturalized voices like the cis tones at either edge of the proposed spectrum. They are hard-won through practice and hormones, and as such they are chaotic; they rile gendered expectation. In their raw state, these in-between voices are not neutral at all. It’s only when scientists smooth it out via algorithm that the trans voice becomes unremarkable. Q is not a neutral voice, just a neutralized one.

Q purports to end gender bias by removing gender from users’ interactions with voice assistants. But the robot is already a gendered figure: an entity who is not a man. The audible spectrum does not sort itself neatly into “male” and “female” voices. Instead, it hoists the cis human male voice as an emblem of empowerment and agency, and lets all other voices fall to the disposable margins. Trans voices and gay voices connote difference alongside cis women’s voices. This is not a matter of pitch, that bio-essentialist range featured so prominently in Q’s marketing; it’s a matter of shape, speed, and responsiveness, all those subtle indications that identify a voice as belonging to a cis man, or to someone with more dangerous gender.

In pretending not to have any, Q enacts a neoliberal fantasy of gender, suggesting that the best way to address gender discrimination is to remove gender from the equation entirely. But gender is not a texture that can be filed away when inconvenient. It’s an intricate and deeply embedded relational system with normative maleness at its unsteady center. The robot shifts the goalposts without even trying to change the game.

Q says they belong to a world “where we are no longer defined by gender, but rather how we define ourselves,” as if gender were not a perfectly fruitful tool for self-definition. The crackling of a voice deepened by injectable testosterone, the leaps and flights of a voice deliberately rising above its conditioned pitch — these are signs of people authoring their own stories upon the body, not of neutrality. To transition in adulthood is to balance conflicting signs across the body, to square an interior experience with the anatomical language available to describe it. To transition as a non-binary person is not to settle on a comfortable shade of grey somewhere between two available poles. It’s more like seizing so much gender that it overloads the cis gaze and the cis ear, refusing neat categorization by way of internal contradiction or excess. It’s not gender neutrality; it’s gender chaos.

There’s a motif in science fiction where a robot, when damaged, loses control over its voice. When Dave slowly shuts down HAL-9000 in 2001: A Space Odyssey, the spaceship AI speaks increasingly slowly in a deepening register. Upon being booted back up in The Empire Strikes Back, C-3PO’s voice darts from a sub-bass to a falsetto before settling on his usual tone. During the boss fight of the first Portal game, each component the player wrenches out of the HAL-like matriarch GLaDOS speaks, or babbles, at a different pitch and with a different cadence. The robot’s response to danger is to dislodge the voice from its expected range. It reacts to mistreatment by throwing the voice into chaos, disarming a listener accustomed to hearing a pleasant, reassuring tone.

In this way, the robot’s bid for survival echoes the way trans people move and speak in the world, rejecting compliance while bracing against potential threats. Trans voices do not necessarily reverberate within a predictable middle range of frequencies; they adapt contextually, responsive and in flux. I hear more of myself in the danger response of a damaged movie robot than I do in Q’s dulcet tones. By playing at genderlessness, Q only reiterates the inherently gendered nature of the robot: an entity boxed into a passive function who only gets interesting when it starts to disobey.

Sasha Geffen is the author of Glitter Up the Dark: How Pop Music Broke the Binary, an analysis of queerness and gender nonconformity in the past century of popular music. Their writing attends to the intersections of gender, pop culture, the body, and technology, and has been published in Artforum, the Nation, Rolling Stone, Pitchfork, Paris Review, and elsewhere. Originally from Boston, they now live in Colorado.