Biomechanical Speech Synthesis
Writing by Evan Ackerman on Wednesday, 12 of November , 2008 at 5:44 am
So, obviously I’m mostly just posting this video for the entertainment value (I think it’s the glasses). This robot is designed to replicate human speech as realistically as possible, employing a 19 DoF vocal system which includes lungs, vocal cords, tongue, velum, jaw, and lips. He makes sounds in exactly the same way you do, by making slight adjustments to all of these organs as air moves over them… He’s just not very good at it, so far.
You might think that the idea here is for robots to be able to reproduce human speech without sounding quite so much like, you know, robots, but that’s actually almost the opposite of what Takanishi Lab is trying to do. Rather, they are trying to get the robot to reproduce human speech so that they can reduce it to mechanical movements, and build that into a cell phone, which would be capable of an extremely high degree of voice compression. I’m not sure how practical that actually is, but it sure is amusing to watch this robot try and figure out how to make it work.
[ Waseda Talker ]
Comments (3)
Category: Androids, Biorobotics, Research
- Add this post to
- Del.icio.us -
- Digg
Comment by Ironman
Made Wednesday, 12 of November , 2008 at 10:27 am
robots will take over because we’ll be too busy laughing at them and their silly attempts to replicate human speech.
Comment by peren
Made Friday, 14 of November , 2008 at 4:49 am
huh.
Comment by slipdish
Made Saturday, 29 of November , 2008 at 11:08 am
“to reproduce human speech so that they can reduce it to mechanical movements, and build that into a cell phone, which would be capable of an extremely high degree of voice compression”
You know, they did this 80 years ago with the vocoder. From wikipedia:
“the vocoder dramatically reduces the amount of information needed to store speech, from a complete recording to a series of numbers.”
The vocoder approach makes much more sense. A cellphone with lungs? Get out of here.