How Transgender Women Are Training Their Voices to Sound More Feminine

For transgender women, the quest for the “right” voice used to begin with a trip to the music store.

“You would go to music stores to get a guitar tuner so you could do your homework and figure out and adjust the pitch you were speaking at,” says Lauren, a transgender woman in Washington, D.C. who requests we not use her full name.

After mobile apps became commonplace, people switched to using electronic tuners, she says, but these only provide an absolute indicator of pitch with no voice-specific feedback.

For transgender women, seeking therapy to modulate to a higher, more feminine voice is about more than identity. “There are tangible safety benefits to being able to pass as cis when you need to,” says Natalie Weizenbaum, a transgender woman and software engineer in Seattle. “Beyond that, I want to be the one in control of how people understand me, and, well, I was just getting really fed up with the sound of my own voice.”

Weizenbaum has taken private voice lessons from a speech pathologist, but at $1,000 for 11 sessions, those can be prohibitively expensive. Now, researchers are developing voice-training apps specifically for the transgender population in hopes of making these lessons more accessible.

Speech-language pathologist Kathe Perez launched the first such voice-training app in 2013. Perez was running her private practice in 2000 when she received a call from a transgender woman who wanted help training her voice to sound more feminine. She started receiving so many similar requests that she put together an audio program that sold in 55 countries. The app—called EVA, or the “Exceptional Voice App”—is based on the audio program and charges $4.99 a lesson.

In two years, some 10,000 users—a respectable but not staggering number—have downloaded the app. Though she created versions for both transgender men and women, far more of her customers are women, Perez says, because feminizing a voice tends to be more difficult than training it to sound more masculine. When cisgender males hit puberty, the extra testosterone thickens the vocal chords to produce a lower pitch. For transgender men, taking testosterone creates much of the same effect, so they require fewer lessons to get to their targets.

For transgender women, though, estrogen treatment doesn’t “thin out” the vocal chords and raise a voice’s pitch, making it more necessary to take lessons or, in extreme cases, have vocal surgery. And some of the issue is cultural, adds Perez: “As a society, we are more apt to overlook a soft-sounding man than we are apt to overlook a very large, masculine-sounding woman.”

These cultural expectations around women and gender have featured prominently in debates over transgender issues. Last year, feminist writer bell hooks criticized transgender actress and "Orange is the New Black" star Laverne Cox for conforming to “stereotypical” ideals of female beauty. Femme coaches who work with transgender clients readily admit that their expertise involves conforming to gender stereotypes. Two years after EVA’s launch, these questions are no less salient when it comes to whether voice training teaches transgender women to speak in a specific, stereotypical manner.

Tools like EVA have specific voice targets with which women can practice. On EVA’s pitch lessons, for example, the app plays a note and the user tries to match the note when singing it into the phone. She then receives a score based on her accuracy.

EVA’s strength is the specific, quantifiable feedback it gives, but this doesn’t mean it’s training everyone to achieve the same female voice, according to Perez. “The human voice has been very well-studied, so we do have parameters and general guidelines of what the characteristics of a female voice are,” she says.

We know, for instance, that the pitch of most female voices hovers about 200 hertz, a measurement of sound wave frequency, though there is natural variation given women’s height and age. Perez built the app to be pitched around 220 hertz, with some wiggle room on each end. If someone’s pitch hits anywhere between 196 hertz and 246 hertz—two semitones above and below 220—she will receive a perfect score. A 22-year-old woman who is 5’6”and a 50-year-old woman who is 5’10” are limited in which notes they can hit by both age and physicality, but as long as they are both within the range, they will both receive 100 percent accuracy. The app averages the results from three different tries, with any score above 80 percent as passing.

EVA provides guidelines, Perez says, but it simply cannot give everyone identical voices even if everyone breathes the same and hits the target range. “A person’s voice is so individual and not just about these numbers—do we uptalk, do we have a darker sound because we’re larger, a bit of a downswing because we’re older?” she says. “All of that ends up coming through.”

The app currently provides lessons in fundamentals, such as breathing style, and pitch, all based on existing language research. The next set of courses, which Perez is still developing, will be about resonance, or the vocal quality that makes a voice sound brighter or darker. This is one of the more difficult aspects for clients to master, says Perez.

Christie Block, a speech-language pathologist who runs the New York Speech and Voice Lab and has coached transgender clients, says the primary tools in her own sessions are unavailable in app form. She uses computer software to monitor her clients’ progress, because it gives visual feedback for continuous speech, whereas most mobile apps can only give feedback for one sustained note. Block praised EVA for making voice training accessible to far more people, but notes that much of voice training involves teaching speech patterns, which an app cannot cover.

“It’s a myth to think that voice training for trans people is just perpetuating stereotypes, but we are definitely dealing with cultural norms like word choice and intonation,” says Block, who refers to “masculine” and “feminine” voices instead of “male” and “female,” because she also works with genderqueer clients. “It’s about helping people understand what the norms are and how to work around them and find the right combination of patterns that make it congruent with their identity and within the biological constraints that they have.”

Soon, EVA won’t be the only one in this field. Alex Ahmed, a doctoral candidate at Northeastern University recently won a National Science Foundation grant to create a voice-training app that “doesn’t conform to a gender binary.”

Ahmed is currently awaiting institutional approval from Northeastern to conduct interviews with trans women to develop her own app as part of her doctoral research. “Personally, as a trans person I know that having a very gendered and very binary interface turns me off, because it presents this idea that there is just masculine and feminine,” she says. “My idea is that people should be able to use my app to further their own goals, which may push them toward different locations on the masculine-feminine spectrum.”

It’s still early in the process—Ahmed says her app wouldn’t be out for another year at least—but she has some ideas for how a more genderqueer voice-training app could work. For instance, there could be customizable voices built in that demonstrate how changing any one factor, such as pitch or inflection, while keeping the other ones constant would affect the sound. This could help people play around with voice training without telling them how close they are to a “female” or “male” voice, Ahmed adds.

She’s also thinking about whether to include more controversial “cultural” features in the app, such as uptalk—the much-criticized (for women, at least) tendency to pitch a voice higher at the end of a sentence. “It’s a very gendered criticism, but I do think that there is value in having as many options as possible, and that’s something that has been observed in the literature,” she says. “I’m not ruling anything out at this point.”

Weizenbaum, the software developer who took private lessons, used tuning apps like DaTuner Pro and Voice Analyst, but describes her learning process as “20 percent biological,” or about pitch and the way her mouth moved, and “80 percent cultural,” meaning it was learning about different speech patterns and how quickly to speak or how to move her voice around. She says, “There was a while when I was paying super-close attention to how people spoke to learn how to sound more emotive, and I became hyper-sensitized to voices in this particular aspect so that every time I heard men talk, I was just stunned at how little pitch variation there was.”

Though she has friends who have had great results with EVA and online training videos, she decided to pay for live feedback during private lessons. “I learn better that way in things I am not confident in, and I was very much not confident about my voice,” she says. She now reports that when talking on the phone strangers identify her as female and use female pronouns 100 percent of the time.

Lauren, the D.C. woman, once took private lessons and now uses EVA about four times a week to keep up with exercises and maintain her voice.

“This is a very long process, but I’m looking forward to all the rest of the modules on EVA, and I’m excited,” she says. “There’s more to learn, and so many more people will be able to learn too.”

Angela Chen | | READ MORE

Angela Chen is a journalist specializing in bioethics, asexuality, technology and education. Her work has appeared in The Wall Street Journal, Guardian US, the Chronicle of Higher Education, Pacific Standard, Guernica Magazine and other publications.