i turned myself into an utau
2023年6月20日
I discovered Vocaloid music on YouTube when I was in 6th grade, which led to a memorable school day trip when I put four (4) songs on repeat for what may have been at least four hours that day. I recall Miku, Len, Rin, Kaito and Luka being the popular Vocaloids at the time.
When I learned that anyone could turn themself into a Vocaloid, I of course had to see what that was about, even if it was 8-9 years later as an adult. In some capacity I considered it an academic exercise partially motivated by a class I took at UChicago. But to be honest I was kinda bored during summer break and was curious if I could make it happen.
To be precise, I didn’t turn myself into a Vocaloid – because Vocaloid is proprietary. Rather, I recorded my own voice to build a Japanese consonant-vowel (CV) voice bank which can then be used in a vocal synthesizer software called UTAU (歌う).
As someone who know much Japanese, after this setup now my voice can be used to “sing” Japanese.
Honestly I did not figure out how to actually use the software to create lyrics / music / a song and I definitely did not know how to do tuning either (i.e. make the resulting singing voice sound more natural), but luckily there are many creators of UTAU Sequence Text (UST) files on YouTube which allow you to use your personal voicebank to synthesize a singing audio file. After combining the synthesized output with the proper backing track, you now have a UTAU cover.
I tried finding one that would work with my voicebank, as there are more complex voicebank styles. Of course it had to be Bad Apple.
Maybe I will one day revisit this project to create my own song with my voicebank, or maybe I can try out some of the more modern stuff. But for now this was a memorable short project revisiting the joys of middle school.
Disclaimer: All credit for music, music video and UST goes to the original creators, this is a derivative work. Specifically, the voicebank was developed myself.
Citations: