天美影院

Ways of Knowing

听

The World According to Sound

听

Season 2, Episode 6

听

Sound Studies

 

[instrumental music plays]

 

Siri voice: Here鈥檚 an answer from Wikipedia.org

 

Chris Hoff: This is the voice of Siri, Apple鈥檚 virtual assistant.

 

Siri: A virtual assistant (VA) is a software agent that can perform a range of tasks or services for a user based on user input such as commands or questions, including verbal ones.

 

Hoff: It can read directions to you, play music, make phone calls, set alarms, send texts and answer any questions for you that you usually use Google for.

 

Sam Harnett: Siri, what鈥檚 Steph Curry鈥檚 free throw percentage?

 

Siri: Stephen Curry has a free throw percentage of 93.3 this season in the NBA.

 

CH: When Siri was introduced in 2011, the only American English voice available was a middle-aged white woman.

 

Siri: I didn鈥檛 get that, could you try again?

 

CH: Fourteen years later, a majority of virtual assistants still sound like white women.

 

[montage of female voice assistants speaking]

 

Golden Owens: Everyone sounds different. And you can learn a lot about a person from how they sound, but you can also learn a lot about a technology from how it sounds.听

 

CH: Golden Marie Owens, assistant professor of cinema and media studies at the 天美影院.

 

GO: Why does she sound so robotic, or why does she sound like a white lady, or why does she sound like a lady in the first place? All of those things can lead to a much broader discovery into things like history, histories of sound and technology, they can lead you into a deeper discussion of race, they can lead you into a deeper discussion of identity and of what it means for things to be chosen deliberately.听

 

You know, it鈥檚 really interesting that the default voice for all of these virtual assistants, at least in the United States, it鈥檚 a white woman. That鈥檚 the standardized default voice unless you change it. Why? And just asking yourself that why can lead you into so many different directions and lead you down a pathway that you may not have expected to go down.

 

CH: One path it took Golden down led to an analysis of servitude in the U.S. After all, these virtual assistants are designed specifically for just that: to do things for us, to serve us. They鈥檙e essentially virtual servants. The history of servitude in the U.S. is a long one, and slavery plays a major part in it.听

 

GO: On the surface, it feels like something that鈥檚 a complete shift because we have these white women鈥檚 voices. But when we think about what it was historically that led to these ideas of what we want in a servant anyways, there was this idea of comfort, there was this idea of something you can have power and control over. In many ways, that also applies to whiteness, but it also is very haunted by ideas of Blackness. And so there鈥檚 a way that you can鈥檛 look at these intelligent assistants as service-providing entities without thinking about where the idea of service came from in the first place.

 

[instrumental music plays]

 

CH: Golden has studied how the way people interact with the white, female-sounding virtual assistants resembles the way people spoke to Black slaves. She began her research after watching how Amazon marketed its virtual assistant back in 2014.

 

GO: It was watching the very first commercial for the Amazon Echo and going, 鈥淥h, there鈥檚 a weird comparison there.鈥� For reference, it鈥檚 a commercial that is about a nuclear family, this white nuclear family, and this little girl is describing all the things that the Amazon Echo can do. It鈥檚 2014, it鈥檚 brand new, and at the end she goes, 鈥淲ith all the things Echo can do, it鈥檚 really become part of the family.鈥� And my brain immediately went, 鈥淭hat is very specific language.鈥� Because that is language that has often been used to describe servants, especially Black servants, as 鈥減art of the family, just like one of the family,鈥� type of thing.听

 

And that essentially sent me on a rabbit hole of like, how much else is Blackness intertwined with the way we think about these virtual assistants? Amazon鈥檚 design guide for years had these things that said: Be adaptable, be relatable, don鈥檛 talk too much, don’t talk too loud, respond to people how they wish to be responded to. All these very specific sort of guidelines for programmers that felt like master-servant language. They felt a lot like the sorts of codes that used to be for how to behave as a proper servant and how to behave as a proper employer. And so for me, it felt like there鈥檚 this intersection of Blackness and technology that is sort of being swept under the rug because they can help us out in our houses, they can help us out in our work, they can do things for us we don鈥檛 want to do, but even that has historical ties to why servants have existed and why slavery existed.

 

[instrumental music plays]

 

CH: Choosing to make the voices of these virtual assistants sound like white women helps obscure those historical ties. Even though you are speaking with these virtual assistants in a similar way to servants of the past, they don鈥檛 sound like the servants of the past. They sound like something new, disconnected from the history of servitude. A white female voice has its own cultural associations. Not because of its objective qualities, but because of how the voice has been racialized. In America, the voice, like the physical characteristics of skin color, hair texture and facial features, was racialized during slavery. People identified and categorized each other based on sound just as much as appearance 鈥� and they still do today.

 

GO: How people sort of hear race ties back into a history of how voices have been racialized throughout history. And that really in the U.S. dates back to the Antebellum era when, Jennifer Lynn Stoever writes in her book 鈥淭he Sonic Color Line,鈥� that there were enslavers essentially that could no longer tell visibly the differences between themselves and their enslaved because of so much assault, basically, and so much race mixing. And so, the sort of workaround for what we can鈥檛 tell visually 鈥� who鈥檚 Black and who鈥檚 not 鈥� is we can tell sonically. So that鈥檚 when we started creating all these definitions of what made a Black voice and what made a white voice. And so the white voice was considered to be clear, calm, controlled, high, but also sort of low energy, in some ways. And Black voices were considered to be fast, loud, coarse, rough and more emotional than white voices.

 

CH: There is no way to design a voice for a digital technology that avoids biases about the way someone speaks. There is no such thing as a 鈥渘eutral鈥� voice. When designing a product, attention is obviously paid to how it works and what it looks like. But just as much thought goes into how the product should sound.听

 

GO: Sound is sort of designed to be something we don鈥檛 think about as much. Especially within a media studies standpoint, there鈥檚 a huge emphasis on the visual, which makes sense. We’ve got movies, we鈥檝e got TV, we鈥檝e got streaming. We鈥檝e got all of these different things. We鈥檝e got VR now. But the sonic and the visual are often working together in a very specific way. In some ways, you can鈥檛 fully understand the visual unless you also understand the sound.

 

CH: In our visually dominated culture, sound is often neglected. We are far less practiced at paying attention to what we hear as opposed to what we see. Sound studies aims to draw attention to this disparity, and recenter the importance of the auditory. Vision may be the hegemonic sense, but there is much to learn if we shift our focus to the ears instead of the eyes.

 

CH: Here鈥檚 five texts that will help you learn more about sound studies as a way of knowing.

 

鈥淭he Sonic Color Line: Race and the Cultural Politics of Listening,鈥� by Jennifer Lynn Stoever

 

CH: Stover explores the relationship between race and sound in the U.S. For her, ideologies of white supremacy are dependent on what we hear 鈥撯�� not just what we see.

 

鈥淗ow Do Voices Become Gendered,鈥� by David Azul

 

CH: This essay challenges the assumption that the acoustic properties of the human voice are determined biologically.听

 

鈥淭he Race of Sound: Listening, Timbre, and Vocality in African American Music,鈥� by Nina Sun Eidsheim

 

CH: Eidsheim studies singers Billie Holiday, Marian Anderson, and Jimmy Scott to show how listeners measure race through the vocal timbres of their voices.听

 

鈥淢ultivocality,鈥� by Katherine Meizel

 

CH: Just like identity, vocality 鈥撯�� how one sounds 鈥撯�� is fluid. Meizel looks at singers throughout history who have reinvented their identities by engaging in what she calls 鈥渕ultivocality.鈥�

 

鈥淭he Possessive Investment in Whiteness: how white people profit from identity politics,鈥� by George Lipsitz

 

CH: A foundational work on the forces that encourage white people not only to keep the status quo, but to invest in structural forms of racial discrimination, or what Lipsitz calls 鈥渨hiteness.鈥�

 

CREDITS

 

Ways of Knowing is a production of The World According to Sound. This season is about the different interpretative and analytical methods in the humanities. It was made in collaboration with the 天美影院 and its College of Arts & Sciences. All the interviews with 天美影院faculty were conducted on campus in Seattle. Music provided by Ketsa, Human Gazpacho, Graffiti Mechanism, Serge Quadrado, Bio Unit, and our friends, Matmos.

 

The World According to Sound is made by Chris Hoff and Sam Harnett.

 

END