I’m not well-educated on the subject (maybe Short King is) but from what I understand, each engine interprets input in it’s own unique (and trade secret-y) way, and that really, really matters. The effect is best demonstrated by “multiple engine” releases such as Gumi V6/SynthV*, where the “input” is likely the same despite the engines’ outputs sounding quite different, even before/without manual tuning. *Gumi is far from the first or only example, but being Gumi, she has easily-searchable comparison videos, so “Megpoid, I choose you.” Sorry if my infodump is overkill.
@@wrong_thyme Hey. Would you be willing to either explain the difference between concatenative synthesis and ai synthesis, or link to something that does? I'd like to learn more, and I'm probably not alone.
@@wellowastaken i unfortunately don't have any links, but i'll try to explain so with concatenative (sometimes called 'standard') synthesis, the voice provider first sings all the sample strings (ex. "aaiauea", one vocaloid vp said they sound like incantations) in a neutral tone of voice, then those samples are basically processed by the synthesizing engine and spat back out with ai synthesis, the voice provider actually sings how they would naturally, and then an ai 'learns' how to imitate their voice (i'm not sure how exactly but that's the gist of it). that's also why ai voicebanks can sing in languages they weren't recorded in (with data from other voicers), but concatenative ones are stuck only using phonemes that were actually recorded. //i think i explained it well, but if anyone who knows more finds something wrong feel free to correct me
She sounds childish, a bit boyish, soft and her pronunciation feels like she speaks with her mouth half closed, so it reminds me of little british boy Oliver V3
@@dummyboi_ there's a Leaked ver of Anri vocaloid in a EARLY development Is not finish and is just japanese For some reason many people triggers when you talk about Vivi