Does Tellme have their own speech recognition tech ? They may have licensed it from someone else. They are more known as a telephony service company. It's one of those phone response service providers.
The doesn't change the nature of the tech or what Molyneux is talking about. He states at the beginning that they use tricks but they're tricks that work. So the question is, to what extent has the technolgoy really progressed. Is there an intelligent being there as Molyneux describes, or if it very far from being that? The proof would be if different resopnse can be got from the experience. If at the end of the day they have a set of situations that the player has ot confirm to, it's not really a learning platform but a join-the-dots where you get the experience they wanted. eg. The skimming - you have to show Milo how to skim and you have to have a fun experience that, after you have had to encourage him, he'll reflect upon.
Not at the rate the demoer was speaking in TED.
There's no need to be unfair! I'm thinking a few voices at the same time, does the mic array actually work, something I ask of PSEye too which introduced it and hasn't used it. We've had not a single demo of working voice isolation.
It's not a case of tracking more than one at once, but that a voice should be detectable given background noise of other people talking, where there's other voices are just background noise. eg. You're watching a film and discussing it, and while the discussion carries on the person in charge calls out some instructions to XB360 to stop/rewind/whatever the film. These array mics should be able to isolate that voice and pass the vocal commands on to the interpreter, where a single mic would just get all voices muddled together. It's this voice isolation that hasn't been shown. And indeed, echo cancellation didn't even make an appearance on PSEye when chatting - its mic array seems completely useless. Until someone actually demos working sound isolation, I remain from experience, skeptical.
His fingers flicker in and out as they lie inbetween sampling points, and he's standing twice as close to the camera than typical playing distance. There's enough there to recognise large hand gestures, like open hand, closed fist, and spread fingers, but not enough detail to do individual finger tracking (at normal play distance, though a foot frmo the camera would work if the setup can focus that closely).
Yes.
There's a valid reason to question it, because resolving the human form in a pose other than standing face on is hard, and gets harder the more deviant from that standing pose you get. Slouching on a couch makes it hard to identify key points in a depth image, and we've all seen the noise in Kinect skeletons where it's had issues resolving other cases. MS also released a video background removal title that didn't work, and a pitch recognition title in Lips that didn't work.
