For anyone growing up in the last 60 years or so, computer voices are a part of the early media you consume. The computer voice on Star Trek was female. The robot Nanny in the Jetsons was female. Many folks either had or selected female voices for their GPS. Now, for anyone with an iPhone, Siri, another female voice is a button touch or spoken word away. Alexa is another female standard option, and Cortana is named for a (barely dressed) character from the video game Halo. But why are there so many female computer voices, and why in particular do so many assistant type programs have one?
Market Research
The first answer is market research. The companies that have created these voices record many lines with the voice talent in question, and then invite consumers in to listen and decide how well these voices bring out things like sincerity and relatability. Their goal is to make sure that listeners feel comfortable with the voices that are helping them organize their lives. But this is only part of the picture. Where do these preferences expressed by the consumer come from in the first place? And even when the data shows this preference, it’s not so overwhelming as to be a decisive aspect. In a study in this article, women preferred female synthetic voices by a margin of 11%, men clocked in a 14% preference. Given these modest margins, more research is probably needed to discover how universal the preference is, and if there’s any change as AI voices have gotten smoother and more nuanced.
The impact of television and movies on modern culture cannot be overstated, and many people who are using this technology now grew up with the evil computer voice of HAL 9000. Even the tech companies joke about avoiding the smooth creepyness of this character. “Bajarin, the Silicon Valley analyst, believes that more computerized voices would be masculine if not for the associations with HAL, whose malicious intent in the 1968 Stanley Kubrick film was made even creepier by his soothing tone. “A lot of tech companies stayed away from the male voice because of HAL,” he said. “I’ve heard that theory tossed around multiple times.” Another element in the preference for the ‘helper voice’ being female can also be traced back to the fact that most telephone operators over the decades have been women.
Female vs. Male Voices
Google has an interesting bit of history behind it’s earlier generation of female voices. At the time the company began its signature voice, they originally intended to create both a male and female option. However, an engineer at Google says, “…He explains that part of Google’s older text-to-speech system, which stitched pieces of audio together from recordings, used a speech recognition algorithm that would place markers in different places in sentences to teach the system where sounds and words began and ended.” This system was trained almost exclusively on female voices, and the engineers had a more difficult time setting the markers for male voices. This led to the function of the earlier Google assistant working better with female voices. The same engineer also voiced the opinion that female voices are easier to hear because of the higher pitch. Other studies have refuted this idea, but it’s given in some places as an explanation as to why the voice assistants are so often female. With recent developments in technology, Google has since upgraded it’s system and added 11 different options for the type of voice you can hear.
But what can you draw from all these factors? Is having mostly female assistants kind of sexist? Not according to this professor. “Not necessarily, said Rebecca Zorach, director of the Social Media Project at the University of Chicago’s Center for the Study of Gender and Sexuality.
“I think they have to be understood in a broader context in which they’re one small piece,” she wrote in an e-mail to CNN. “Voices intended to convey authority (such as voice-over narration in films) tend to be male. So yes, probably these compliant female robot voices reinforce gender stereotypes, not just because they serve the user but because the technology itself is about communication and relationships (areas that women are presumed to be good at).I wouldn’t automatically claim any sexism in individual companies’ choices, though. Most such decisions are probably the result of market research, so they may be reflecting gender stereotypes that already exist in the general public.” Zorach listened to some sound clips of Siri online, then e-mailed back again. “What’s interesting to me is how they seem to intentionally make her speech sound artificial – they could choose to make her speech more seamless and human-like, but they choose instead to highlight the technology,” she said. “That makes you aware of how high-tech your gadget is.”
Societal Influences
Obviously there are many factors involved in how these voices are put together. But it’s important to consider the sociological and historical factors when evaluating where the preferences come from. Objectively speaking, choose whatever voice you enjoy the sound of, of course. However, there is more than just market research involved here. When it comes to voices of authority, the chosen voice is often male. This works for announcer voices, movie trailers, and other similar applications. For many decades, most radio announcers were male also. Early audio recording equipment was created in such a way that female voices were often distorted or shrill, which led to the preference for male announcers. Women were coached to drop the pitch of their voice, and to speak without emotion, which is why many female leaders or voice personalities have low, smooth voices. At the time, it was considered to be a psychological and biological issue contained in women themselves, rather than being a factor particularly related to recording equipment. In addition, due to the stereotype that women spoke softly, male engineers would often raise the levels when a woman came to the microphone, further worsening the distortion and unpleasant sound. Doubtless this contributes to the roles given to female voices, in that they fit the general stereotype of ‘womanly’ roles.
In addition to these other factors, it’s also important to note that it’s not universal. Many countries do have default female AI voices, but the UK, Germany, and a number of other countries have a default male voice. In particular, male German drivers complained to BMW in the 1990’s about receiving directions from a woman, and demanded a male option.
So when you put all these factors together, it’s fair to say that the preference for female voices is extremely common, but not universal, and some of it has been led by the manner of technology creation and understanding. Other parts have been influenced by the decades of radio and television voice gender choices. We would probably have different preferences if women had been given a fair shake in those early days. But there’s culture too, since voice preferences aren’t the same from country to country. So is it sexist? Probably, but the most important thing for the AI assistant that you prefer is that you prefer their voice, and there will only be more options and nuances to create the robot helper of your dreams as time marches on.