Gekkayo posts raw VOCALOID3 comparisons
In time with the release of the magazine “Gekkayo VOCALOID fan Vol. 4” (a magazine/book hybrid, i.e. “mook”, containing tutorial, interviews, sheet music and other info), a collection of “raw” VOCALOID3 audio from a plethora of sound banks have been uploaded to Nico Nico Douga as well. These are all of the same short song where the vocals were created by directly inputting the notes and syllables into the VOCALOID3 software with no tweaking or tuning involved. Outside of mixing with the backing instruments, the audio for the vocals has not been processed in any other way (e.g. via a compressor or EQ) after synthesis. The sound banks demonstrated were: IA, CUL, Mew, SeeU, VY1V3, Tone Rion, and Yuzuki Yukari. These samples accompany a section in the book that compares these seven sound banks.
Ooh this is very nice to have a comparison like this. Thanks to the Vocaloid 3 engine, and probably a better technique to record the samples, they all sound very smooth and clear. Just a couple of things I noticed:
1 – Yukari, at the beginning of the song, tends to clip some of her consonants. Probably a simple fix but none of the other voice banks did that.
2 – SeeU seems to have a bit of accent in some places, which I guess I expected.
Other than that, a user really can’t go wrong with any of these Vocaloids. Just a matter of personal preference at this point.
Youtube mirror please! (I’ll love you eternally)
… aha, I’m sorry, but when I read “mook”, I thought of the Timon and Pumbaa cartoon. Fun memories.
Anyways, this is interesting. Maybe someone will use these raw samples to develop better UTAU samples? I mean, known how the plain Vocaloid samples sound would be a really good reference point in figuring out how good an UTAU’s samples should be… but that’s just a thought.