Engineers and scientists at Kagawa University in Japan have been working on an interesting device: a synthetic mouth, tongue and windpipe that can physically produce human speech. The artificial production of human speech has been around (and been creepy) for a long time, but the selling point here is that, unlike a Speak & Spell, or a woefully unnatural text-to-speech program, this synthetic mouth, windpipe and set of vocal chords can make noises without using speakers or anything digital. It makes noises the good old fashioned way.
When it’s all set up and calibrated, the mouth makes sounds through lip movement and air flow that are then picked up by a nearby microphone, analyzed by a computer that determines the necessary pitch correction, and that information is sent back to the mechanical rig operating the throat. This way, the synth mouth is eventually “tuned” and eventually sounds like a human.The actual applications of this technology, beyond hanging around and being uncanny, are probably pretty far off, but it’s certainly and interesting step in vocal technology. I, for one, would prefer more believable digital text-to-speech technology over weird synth flesh tubes that can moan and do things that are creepy when weird synth flesh tubes do them, but hey, maybe some day those weird synth flesh tubes will be able to argue their case and that’ll shut me right up.
More pictures, a diagram and a video after the jump.
(via Trends in Japan)
