RealPlayer | Lo QuickTime | Lo Windows Media | Lo
Harry Gohs - talks us through Cantor
Showing for the first time was the Virsyn vocal synthesizer Cantor. Based around a sythesis engine that allows control over human phenomes (mouth sounds), pitching and noise, Cantor doesn't use any samples.
The main screen allows you to enter text to sing back when playing notes from a MIDI keyboard. The results are pretty good, and this is only scratching the surface. It's possible to edit the spectrums for the vowels and constonants of the voice, plus the spectrum and level of the noise content, allowing dynamic modulation of the breathiness of the voice. For instance, by hitting the notes harder, it's possible to go from a soft breathy vocal sound to a more forceful aggressive tone - this makes the level of realism pretty darned good.
Cantor doesn't just work on true realism, but can be set to make much weirder and more alien sounds. Currently, Cantor speaks only English, but ther will be further language packs coming, or you can create your own phenome sets for alien languages.
The important thing to remember with Cantor is that it performs this synthesis in real-time from a MIDI input.
8 part vocal synthesis engine
Sadly, I don't have a screen grab available, but we did shoot an interesting video piece with Cantors creator Harry Gohs, which will be in place shortly.
Cantor is available in standalone mode, VST2, AudioUnit, ReWire and RTAS. Expected shipping May 2004 +/- $300