Spanish voicebank development

 

    As many of you may know, I've been working on a Spanish voicebank for Suu. I'd like to talk more about it... 

    I began working on this voicebank since some months ago (maybe May, I think...). This is my first time developing a non-japanese voice library and, dude.. this is being such a journey; I tried a whole different phoneme recording system (nJokis VCCV) along with a long ass reclist, like... 150+ recording lines. And things become even more fun because this is a multipitch voicebank!! I've recorded 4 pitchsets (E3, G3, C4 and F4). So... approximately more than 150 recordings per each one. Let's not talk about the oto-ini... The development of this voicebank is taking longer than expected due to many factors (daily power outages, I wasn't alone at home most of the time so I wasn't able to record freely, I was busy at the family business... etc.), I'm trying my best though. I've set a release date, and it will be... JANUARY 3RD, 2027!! So, please look forward to it.

 

 

 

    The folder's size is roughly 300MB... holy shit!! Well, I kinda expected it; Spanish is a huge and complex language. Besides this isn't even the 50% of the size of a VOCALOID3 voicebank. LOL

    Some time after I made Suu's first voicebank, I dreamed with the possibility of him singing in Spanish; my native language. I'm finally making it come true. Since I started developing this new voicebank, I imagined Suu singing a lot of songs by hispanic/latin musicians I like; mostly rock, or even very popular songs that are part of Latin American culture. This led me to look for songs I could make him sing during the testing process of the voicebank. 

    One day, I looked for some .ust/.vsqx files on YouTube, but... I ended up feeling a bit disappointed because of the very little amount of sequence files of Spanish songs that were available for download in comparison to Japanese and English songs. Not to mention that most of these sequences were Spanish adaptations of songs by Japanese musicians. I mean... it's kinda sad to find out there is a very small diffusion of hispanic/latin music despite a huge portion of the vocal synthesizer community is conformed of Spanish speakers.

    So... I had the idea of making my own .ust files of songs I like!! I suck at making music sequences in general, but I'm trying my best (2). This also led me to look for songs and musicians I didn't know nor hadn't listened to before. It was also a good opportunity (and excuse) to even learn more about my country's rock music culture. 

    I ended up discovering a few albums and songs I hadn't listened to by one of my favorite bands (Frisbee by Caramelos De Cianuro, especially. This album is so cool). And, dude... I re-discovered a band I had found about some years ago but hadn't paid attention to... until now. I decided to listen to Sentimiento Muerto's entire discography and learning more about them. Their sound is MY kind of music... and I'm still obsessed with a lot of their songs. Besides, their style fits a lot with Suu's Spanish voicebank melancholic and transgressor concept. (I might talk about this in another post...) I'm already making a cover of one of their songs. :)

    ...ANYWAYS, I got carried off a little. Let's continue talking about the voicebank. Yes. The voicebank. I'm currently setting the oto.ini files up... this is going to take some time. Most entries from the base pitchset (G3) are done. Even though, I really like how this voicebank is coming out. It also resembles my real voice more than the rest of Suu's Japanese voice libraries do.

  

 

 

    So... here's a little test of this voicebank, I liked this one a lot. I must lock in and finish it soon. Suu has to sing a lot!!

~ Jakisalt