I wanna sing someday/いつか歌いたい
A recollection of memories: Karaoke, chorus, and vocal synthesis software. And, why can't a transfem sing?
One spring day last year, I took the train to the neighboring city to see Y's band performance. It was a show put up by the anime and video game music club of Y's university. Since joining the club months earlier, Y had been filling her life with rehearsals and hangouts with fellow bandmates. As a huge fan of band-themed fiction, I couldn't let go of the chance to witness a slice from the real-life version of it.
After a sudden afternoon rainstorm passed and the night began to fall, groups of students (and I) filled the seats in the auditorium, and the bands started to play. Of course no one on stage was professional, but the performance was impressive nonetheless. The crowds were cheering, the performers were clearly enjoying themselves, and I was ingesting fresh youth enthusiasm.
During intermission, they left some background music playing. By the stage, I recognized R, Y's partner, who was a slightly androgynous white boy with long curly hair. As I watched him from afar, he sat down at the edge of the empty stage, picked up a microphone, and began to sing to the music, karaoke style. It was a song from LoveLive!, one of my favorite anime franchises. At that moment, he was singing casually and expressively, as if there weren't a roaming crowd in the room potentially watching. After one song ended, he just followed the music to the next. How free!
"I wanna sing like this someday." This was the thought that hit me next.
•
"Do you like Karaoke?" My SLP[1] asked during our first session.
"Not really", I answered.
"I was asking because another of my Chinese trans girl clients loves karaoke, so for her we had to do a lot more work on the singing front." She explained, "So that's good."
But my answer was grossly oversimplified.
I loved singing when I was a kid. As a result of my parents' insistence that I learn to play the piano, I had a great sense of pitch and could grasp sheet music quickly. My parents even sent me to singing lessons a few times, although I gave up soon due to my peculiar music taste (mostly English and Japanese, which bodes ill with Chinese musical classrooms).
When I started middle school, I got the freedom to hang out with friends without my parents present. And I quickly became fond of the newly installed karaoke boxes in the local shopping mall. They were basically soundproof phone booths with karaoke machines installed inside. Contrary to the more traditional KTV[2] establishments, they were much cheaper, did not smell like tobacco, and did not require ID to use, thus perfect for teenagers like me. My friends and I frequented the mall on weekends and holidays, and we would usually pop into a karaoke box to sing a few songs after a long, bored stroll.
Days passed by and when I was slowly learning to sing anime songs using Romaji[3] lyrics, puberty sneaked up on me. One day I was singing in the karaoke box as usual, and one of my friends decided to record a video of me singing "Masayume chasing" (in hindsight, a very challenging song). The video was then sent into a larger friend group chat. Soon, laughs and ridicule filled my chat inbox.
"Am I really that bad?" With some defiance I opened the video. And that was the moment when I realized, painfully, that my voice had defied my best efforts to control, turning lower and more creaky. My body had changed in a terrifying way, and singing was never the same.
Deeply humiliated, I stopped going to karaoke for years. Until around the time I graduated from high school, and had found a group of hardcore otaku[4] friends to fool around with.
Over the following summer and winter holidays, we would go to the only KTV in the city that offered an extensive Japanese song library, and stay there for multiple hours. They didn't care how well I can sing, as long as we could all spend the time singing along to the latest and hottest J-POP and anime tracks. It was okay to shout, rap to the lyrics, or only sing the chorus part. We all had our mental complexes, and the dark rooms in the KTV seemed to have become a safe space to escape from the trappings of daily life.
But after the group dispersed to different cities or even countries, the space gradually disappeared, along with my familiarity with singing.
•
Once during my undergrad in China, I was "compelled" to sing. It was the mandatory school-wide chorus competition, and freshmen from each department had to team up to perform in front of the whole school. It was counted as part of the propaganda class credit requirement, so there was no escape.
On one of the chilly autumn nights, after a full day of classes, all students from the math department that I belonged to convened in a large classroom to be assigned to voice types and learn about the assigned song for us. I didn't quite remember which voice type I was: Bass or Baritone? Probably Bass. There were a lot of Basses, and very few Sopranos and Altos, because in my cohort, only 10 percent of math students were women.
The conductor, who was an older student in the department, also had the talent to write the song for us. He was determined to put something "distinctly mathematical" in the song, so one line of the song's melody followed the digits of Pi[5]. As a result, it sounded rancid. Or maybe even well-written music wouldn't sound good on such an imbalanced chorus.
But nonetheless, for weeks we practiced every other workday. The Beijing winter came, and when a hundred male college students got squeezed into a heated room with no airflow, the room turned into a sweaty and smelly sauna. Quite possibly, no one really wanted to sing in the chorus or perform the song, and the group just chugged along following whatever instructions were given by the teachers.
On one night where the rehearsal dragged especially long, I suddenly heard disruptions on the other end of the crowd formation. I turned to look, and saw one of my classmates lying on the ground, shouting, "I don't want to practice anymore, I want to leave and do math homework[6]!" The people standing beside me whispered, "what a weird guy!" I just stood there and watched the situation unfold.
Years later, long after I left the university, I started to make more sense of that night's happenings. Even when lashing out (possibly in an autistic meltdown), he chose the most conformist wording to make himself legible to his environment.
My memory of the day of the chorus performance was blurry. But I did remember I intentionally kept myself silent for a large part of the performance. My mouth moved in sync with others, but no sound would come out of it.
Staying silent could be a form of protest.
Last year, my queer friends would often invite me to join their local "LGBT-friendly" Chinese chorus. "They really need more Bass members." They tried to convince me. I never thought about joining. As long as they still fit voices into pre-determined, gendered types, it is not for me.
•
During the years when I was dismayed about singing myself, I became deeply intrigued by synthesizing singing voices using computer software. The technology was born from Vocaloid, which could use a set of pre-recorded singer libraries to create distinctive vocals. It became so popular that it had spawned a whole new genre of music, and cultural icons like Hatsune Miku. The software that I first learned to use, UTAU, took the idea further and could synthesize a singing voice from any sound material on the Internet. In the hands of infinitely creative netizens, UTAU songs, covers and their accompanied "voice banks" could become memetic hits in the "Human-powered vocaloid"[7] subculture corner of Bilibili[8].
UTAU was hard to learn, and even harder to master. Released in 2008, it had the look of Windows 98 and a rustic, rudimentary feel to it. To build the vocal for a song involved putting the syllables together one by one, according to the pitch line for the melody. To make the concatenation of syllables even remotely resemble a singing voice, one needs to tune many arcane settings to modulate how syllables are connected, after understanding how words in the lyric should be pronounced. Even when all the proper tricks were used, the resulting voice would often still have a low-fidelity sound profile. It was both UTAU's fatal flaw and where its attractiveness came from.
I remember the awe I had when I first tested out Synthesizer V, the newer software, during my freshman year. As a result of adding neural network based voice synthesis, it was so easy to use, and even for a novice like me, the resulting vocal sounded somewhat natural. It felt comforting to me, that even if I had lost my singing voice, the avatars in the software would sing in place of me, channeling my feelings.
That was before the explosion of Generative AI, and nowadays generating human-sounding voices is hardly a novelty anymore. In the web platform that I use to record podcasts, I have a list of "AI speakers" to pick from. A quick search would reveal a proliferation of AI models that could convert one's voice to match a desired character. However, most Vocaloid creators stayed on older generations of software, where their artistic direction still matters.
Even when I was struggling the most with dysphoria, I never thought about using an AI voice to replace my online self. The feminine AI voices sound impeccably good, but also too stereotypical and sometimes uncanny. I ended up going to voice therapy, which fixed my speaking voice for the most part. But not my voice for singing, due to cost and my limited capacity.
These days, being able to sing again remains only a distant wish of mine. In the unlikely event of me one day regaining the energy of an American college student, I would actually make the effort to learn how to train my singing voice, and sing for nothing but personal joy and artistic pleasure. But for now, I will just be humming my silly little tunes at home.
Thank you for reading it all. You can find all my essays using the tag.
A locally-deployed Gemma 4 LLM was used to check language and grammar issues.
voice therapy professional ↩︎
Asian term for a karaoke place, usually contains multiple rooms, and provides drink service ↩︎
Latin script for Japanese ↩︎
zealous anime or manga fan, sometimes derogatory ↩︎
In Chinese Jianpu, 1 is the middle C, so 3.1415926 became . ↩︎
The original wording, "做题", means "to solve math problems". But in a school context, it often means the problem sets given in a class. ↩︎
Translated from Japanese "人力ボーカロイド" ↩︎
Popular video sharing platform in China ↩︎