I will say the issue with V3 onwards was mostly in Japanese due to the growing problem with a lot of vocals not producing distinct enough vocal results. I call ignroance here sue to basically knowing less about them in general and not much on the langauge O.o I have no comments on Chinese, Korean and Spanish since. A bit of English comments as well (when they weren't laddled with bias). So this is the general consensus on both English and Japanese.
My responses are based on what was said by the Japanese communitity collectively from the V2 era. While English vocals have trouble maintaining quality, they usually are considered better at capturing their provider's traits due to having many ways to capture traits and thus excel note for note in realism over japanese vocalsas they have 8,300+ samples to get traits from. IF we're talking realism too in this, All English vocals are realistic except Sonika from Zero-G, Big Al is voice acted but as far as anyone knows the other PowerFX vocals are not. Big Al and Prima were the best of V2 English vocals. Luka English and Sonika are the worst two English vocals while Sweet Ann and Tonio are the MQ vocals. This lead to V4 era when they overhaulled the recording process to make Vocaloids harder to use in favour of more vocal traits in Japanese voicebanks because too many were ending up too similar overall. Japanese vocals can have realism uissues due to getting traits from about 1,500+ samples. In terms of realism, Gackpo and Gumi have traits of their provider's but weren't quite there, but the Ah Software trio are quite realistic, though Gumi's generally considered of when Japanes vocals became more realistic they had issues. vague, since not every producer got to try most of the vocals and the climate was not so focused. Most vocals were MQ after 2009 in Japanese except for the Appends and VYs as they were HQ. Miku V2, the Kagamines, Gackpo and Gumi are LQ, while Luka was the start of the HQ vocals releases (being the worst of the bunhc), while Ah Software trio was the start of V2 MQ range, with Miki being notable for the first vocal to produce unique enough vocals to fill all 16 tracks and NOT loose quality/tone.
Miriam was just meant to sound like Miriam Stockleigh and was the only one trying to mimic their provider to a tee. Leon and Lola were meant to convey traits of a particular style and were both pretty good, though Lola was worst due to atypical accent results.
Meiko and Kaito have the issue of they were made as "no thrills" vocals and therefore neither were made with vocal traits in mind, this predated Miku basically. V1 era: Meiko is overall worst then Kaito for Japanese, while Leon, Lola and Miriam follow lowest to highest in that order.
0 kommentar(er)
