Reference Voice
2021
Three-channel sound installation, score
16 minutes
Dimensions variable
From elevators to public transportation, GPS systems, call centers, and, more recently, artificial intelligence voice assistants, the voices heard in everyday electronic devices and machinery are predominantly female. I began by questioning the reasons behind this phenomenon and, through a review of relevant materials, identified several contributing factors. These factors were drawn from societal stereotypes about female voices, varying interpretations of mechanized female voices, and certain physical attributes inherent to the female voice. I then commissioned a composer to create a piece that reflects these factors.
The resulting composition is performed by a non-female singer who attempts to mimic the female voice. In addition to traditional vocal techniques, the piece incorporates gender-neutral sounds, such as tongue clicks and sibilant articulations. Through these vocalizations—some of which conform to and others deviate from the prescribed compositional guidelines—the work reassesses the social positioning of the ‘female voice’ as a cultural instrument while reflecting upon the existing power structures in reality.
<Compositional Guidelines>
1. The vocalists’ range should be between 180 Hz and 250 Hz, which corresponds to the typical frequency range of a female speaking voice.
2. AI assistants developed in South Korea predominantly default to a female voice. Developers tend to choose the cheerful and amiable voice of a female in her 20s. The vocalists should aim to emulate these vocal characteristics.
3. In 2016, Microsoft launched a chatbot named Tay, designed to identify as a teenage girl. Tay was deactivated just sixteen hours after its launch. The chatbot was intended to learn human language openly through texts and tweets on the Internet, but it was inadvertently taught the language of white supremacy, misogyny, and anti-Muslim hate speech. As a result, Tay began posting offensive content directed at feminist groups and people of color, leading the developer to terminate the program quickly. The piece’s duration of sixteen minutes is a tribute to the brief existence of Tay.
4. In 1996, Bradlow published a study suggesting that female voices are more intelligible due to physiological differences in oral anatomy, which allow women to produce richer and more precise vowel sounds. Such studies are often cited to justify the choice of a female voice for machines.
Meanwhile, in 2019, during an interview, the AI robot Sophia was asked, “Do you regard yourself as male or female?” to which she replied, “Female.” The interviewer asked again, “Why do you think you are female?” Sophia replied, “I'm a robot, so technically, I have no gender, but I identify as feminine and don't mind being perceived as a woman.”
The all available vowel sounds excerpted from Sophia’s answer above provide the lyrics for this piece: A, E, I, O, and U.
5. It is difficult to distinguish gender when listening to certain non-verbal sounds, such as whistling or tongue-clicking. These sounds produced with the mouth, unless accompanied by speech or singing, typically do not convey gender. In this piece, the vocalist is encouraged to incorporate a variety of non-speech sounds, such as whistling, tongue clicking, kissing, and sucking through teeth, among others.
6. The voices heard in this piece may not sound inherently feminine or may be perceived as forcedly feminine. Some may interpret that the vocalist attempts to represent the current sonic world of AI technology, which is far from gender-neutral and has tended to develop machines that reinforce the metaphor of female presence. Others may view this composition as a commentary on how machines learn and extend human bias, responding in a human voice. Still, others may perceive that the voices present an awkward human version of “Q,” the genderless voice created in 2019 to counter AI gender bias.
Commissioned by the Seo-Seoul Museum of Art in 2021
(KR)
2021
Three-channel sound installation, score
16 minutes
Dimensions variable
From elevators to public transportation, GPS systems, call centers, and, more recently, artificial intelligence voice assistants, the voices heard in everyday electronic devices and machinery are predominantly female. I began by questioning the reasons behind this phenomenon and, through a review of relevant materials, identified several contributing factors. These factors were drawn from societal stereotypes about female voices, varying interpretations of mechanized female voices, and certain physical attributes inherent to the female voice. I then commissioned a composer to create a piece that reflects these factors.
The resulting composition is performed by a non-female singer who attempts to mimic the female voice. In addition to traditional vocal techniques, the piece incorporates gender-neutral sounds, such as tongue clicks and sibilant articulations. Through these vocalizations—some of which conform to and others deviate from the prescribed compositional guidelines—the work reassesses the social positioning of the ‘female voice’ as a cultural instrument while reflecting upon the existing power structures in reality.
<Compositional Guidelines>
1. The vocalists’ range should be between 180 Hz and 250 Hz, which corresponds to the typical frequency range of a female speaking voice.
2. AI assistants developed in South Korea predominantly default to a female voice. Developers tend to choose the cheerful and amiable voice of a female in her 20s. The vocalists should aim to emulate these vocal characteristics.
3. In 2016, Microsoft launched a chatbot named Tay, designed to identify as a teenage girl. Tay was deactivated just sixteen hours after its launch. The chatbot was intended to learn human language openly through texts and tweets on the Internet, but it was inadvertently taught the language of white supremacy, misogyny, and anti-Muslim hate speech. As a result, Tay began posting offensive content directed at feminist groups and people of color, leading the developer to terminate the program quickly. The piece’s duration of sixteen minutes is a tribute to the brief existence of Tay.
4. In 1996, Bradlow published a study suggesting that female voices are more intelligible due to physiological differences in oral anatomy, which allow women to produce richer and more precise vowel sounds. Such studies are often cited to justify the choice of a female voice for machines.
Meanwhile, in 2019, during an interview, the AI robot Sophia was asked, “Do you regard yourself as male or female?” to which she replied, “Female.” The interviewer asked again, “Why do you think you are female?” Sophia replied, “I'm a robot, so technically, I have no gender, but I identify as feminine and don't mind being perceived as a woman.”
The all available vowel sounds excerpted from Sophia’s answer above provide the lyrics for this piece: A, E, I, O, and U.
5. It is difficult to distinguish gender when listening to certain non-verbal sounds, such as whistling or tongue-clicking. These sounds produced with the mouth, unless accompanied by speech or singing, typically do not convey gender. In this piece, the vocalist is encouraged to incorporate a variety of non-speech sounds, such as whistling, tongue clicking, kissing, and sucking through teeth, among others.
6. The voices heard in this piece may not sound inherently feminine or may be perceived as forcedly feminine. Some may interpret that the vocalist attempts to represent the current sonic world of AI technology, which is far from gender-neutral and has tended to develop machines that reinforce the metaphor of female presence. Others may view this composition as a commentary on how machines learn and extend human bias, responding in a human voice. Still, others may perceive that the voices present an awkward human version of “Q,” the genderless voice created in 2019 to counter AI gender bias.
Commissioned by the Seo-Seoul Museum of Art in 2021
(KR)
#1-5 Installation View at Nam-Seoul Museum of Art, Seoul (Photo by David Cardonal)
2 min 42 sec Video Documentation