YAMAHA has released both VY2V3 and VocaListener yesterday, and both are available on the VOCALOID STORE. VY2V3 is available in a 9800 JPY standard package, a 17800 JPY Starter Pack and a 9800 JPY download version on the Japanese store. VocaListener is a plugin for the VOCALOID3 Editor and thus there’s only a download version for 19800 JPY on the Japanese store. The only product available on the English store is the download version of VY2V3, probably owing to the fact that VocaListener does not support English phonemes.
In the days leading up to the release of VY2V3, several official demos were released to whet fans’ appetites:
- Cendrillon (VY2V3+VY1V3 version) by Dios/SignalP
- Tottemo Itai Itagaritai by EZFG
- Aru Futekigousha no Dokuhaku by Kuroda Asin
- Nurebairo by Camelia
As for VocaListener, ITmedia News wrote an article detailing the history of the project and also provided three example songs. Right now, several songs/clips made by users have been uploaded on niconico that use VocaListener: IA covering a fragment of Aozora, in various stages of tuning, various VOCALOIDs speaking (based on VOICEROID+ Yuzuki Yukari), Iroha singing Rera, GUMI singing Cross Road, Luka singing a short clip of Aoi Shakunetsu, IA singing Fade Away, Miku Append Solid attempting to sing an English song, Roundabout and Mew covering Kanashimi Honsen Nihonkai.
The overall workflow for VocaListener appears to be the following. First, an empty song block needs to be selected and the program started inside VOCALOID3 Editor through the JobPlugin menu. VocaListener pops up as its own GUI, complete with waveforms and notes, etc. Once the window is up, the user needs to load the source audio, after which VocaListener automatically analyzes the pitch. Then, the user needs to drag out areas that have singing and type in a string of lyrics that correspond to the lyrics sung. The program will then think and adjust the positioning of the notes, as well as readjusting the pitch and dynamics curves. Often times, the program’s guess for the timing of the lyrics is not correct and need to be adjusted. Other parameters may also be tweaked, including the amount of pitch expression, etc. Everytime there is a change, it’ll take some time for VocaListener to process and redo the pitch and dynamics tweaks. Once the user is satisfied with the tuning, one button closes the window and sends the note, pitch and dynamics data back to VOCALOID3 Editor. There is also the option to load and save VocaListener projects, and launching VocaListener from a track made with VocaListener should bring up the original project associated with it.
VocaListener will require a very clean audio source for it to work well, i.e. a raw audio recording with no reverb, etc. This is because it’ll attempt to match the entire waveform, and effects such as reverb will cause VocaListener to attempt to emulate that with the singing, resulting in very funny results.