The non-standard game interfaces discussion thread (move, voice, vitality, etc.)

patsu said:
I think the phones have pretty bad speech recognition. If they can improve on the basic quality, then it may be worthwhile. SingStar has such song/artist name recognition capability, and it is rather convenient.
Click to expand...
It's easy to recognize songs when you're only selecting from a small-preset list of acceptable songs, it's much harder to apply it to your mp3 collection like the iphone tries to do.
My car has bluetooth voice dial and although i didn't have to train it, it dials perfectly any 10-digit number I speak, one digit at a time.
 
Just wish to share with you this article explaining why voice command is, as my point of view, totally useless : http://robertfortner.posterous.com/the-unrecognized-death-of-speech-recognition

All the buzz about voice driving of Kinect/Xbox as a mediaplayer or anything else will be useless. As explained, you will have fun with it a few times, and totally forgot it because of too much false positive order.

With this article in mind, I am just thinking about Milo... Do you really imagine a 'game' will be more advanced and intelligent than whatever speech recognition we have actually in the market ?
And one thing to keep in mind for US people : all other countries do not speak english. So if english recognition is already very difficult and need a large power of artificial intelligence, imagine if you are using a language as difficult as japanese or french ?
 
I don't know. I'm skeptical, but in the right context it can still beat alternatives. For instance when playing Sing Star and you're both standing holding microphones, then being able to use voice commands can be convenient. My personal experience is also that it works pretty well, both in Singstar and on my iPhone 3GS with OS4 (I never tried it before OS4).

However, it is definitely true that if you just mumble something non-existent, it will always try to find something that resembles that as much as possible anyway. This is partly why the commands on the 360 are always prefaced by 'Xbox ... '

It is definitely healthy to remain skeptical though. I hate using voice commands because I don't like using my voice for these kinds of tasks in the first place.
 
Gon0S said:
Just wish to share with you this article explaining why voice command is, as my point of view, totally useless : http://robertfortner.posterous.com/the-unrecognized-death-of-speech-recognition

All the buzz about voice driving of Kinect/Xbox as a mediaplayer or anything else will be useless. As explained, you will have fun with it a few times, and totally forgot it because of too much false positive order.

With this article in mind, I am just thinking about Milo... Do you really imagine a 'game' will be more advanced and intelligent than whatever speech recognition we have actually in the market ?
And one thing to keep in mind for US people : all other countries do not speak english. So if english recognition is already very difficult and need a large power of artificial intelligence, imagine if you are using a language as difficult as japanese or french ?
Click to expand...

This article might be revelant if we wanted our 360 to take dictation, but we are talking voice control used for a limited set of commands, which is a lot more feasible.
 
dobwal said:
This article might be revelant if we wanted our 360 to take dictation, but we are talking voice control used for a limited set of commands, which is a lot more feasible.
Click to expand...
Yes, indeedy. We're world's apart from computers being able to follow vernacular speech, and TBH I doubt it's possible, but the intention with voice control is using a set of sounds to control events. The system only needs know a certain degree of vocal communication, and we shift the need for intelligence to the human operator. Instead of having to parse, "I want to see a movie," and, "what films are there?" and "Shit, what a day! Put something on," as all being instruction to fire up the media capabilities, the system only need parse, "Xbox, film," or somesuch with the user learning to select the right vocabulary.

The problems then are purely a matter of associating the sounds with the keywords; matters of audio analysis.

Having said that though, the Milo demo was very suggested of natural speech recognition, which if people are believing they'll get, they'll be disappointed!
 
Shifty Geezer said:
Having said that though, the Milo demo was very suggested of natural speech recognition, which if people are believing they'll get, they'll be disappointed!
Click to expand...

I'm personally thinking the Milo demo while allowing a person to use natural speech is like early Infocom and Sierra adventure games.

You could type in sentences if you wanted, but the game was really only parsing for keyworks.

So in the case of Milo for example. A person could say, "Milo do you want to play a game?" And it's quite possible that the program is only parsing "milo", "play" and "game" out of that and ignoring everything else. Additionally to make it easier to parse, it may be setup to only parse words that are distinct and don't have multiple very similar sounding words.

Regards,
SB
 
Yes, that's my guess too. Which'll be good enough to fool most people, alhtough I'm very interested with how it'll work with regional dialects and accents, and foreign pronunciators. I'm thinking Chekov from the latest Star Trek film. ;) This is actually a key point in language congnisance, the fact that a person can not know a word but can learn its meaning (or at least workable approximation) mid-conversation from context, common enough with slang. No reference lexicon is going to cover our ever evolving language and its fashions. When did 'cool' take on the meaning of a particular type of excellence? Any computer parsing language at the time would have been completely thrown. Then again I suppose parents are too, so as long as the computer can grumble about the state of modern language and ask the Yoof culture to explain itself, maybe it could keep in the know. (elloh-el, ohemm-gee, arrohty-effell :mrgreen:)
 
Free form speech may be hard. "My Playstation game blah" may get matched to "Milo play game" too. My old speech recognizer required me to pause when addressing the system:
"Milo" <computer acknowledge after 0.5 second> "play game" <kick off action after 0.5 second>

If the system makes a mistake, you'll need to repeat the sequence. That's why it can be slow.

It's also unclear whether you want to play a song called "Play Game", or you want to play a game. Further choice may be needed.

If the system is expecting input in a known context and format, e.g., waiting for song selection, then the success rate will be higher.

Microsoft has also clarified that Milo is not a game:
http://www.vg247.com/2010/06/29/microsofts-greenberg-milo-is-a-tech-demo-no-plans-to-release-it/

Speaking on ABC TV’s Good Game, Xbox boss Aaron Greenberg said Peter Molyneux’s mildly creepy virtual son probably won’t ever come out to play.

“Milo; he’s safe and sound back in England. No… the Milo Project is something that Lionhead Studios in their labs had developed. Last year we unveiled the Project Natal technology, we showed a bunch of technology demos as part of that,” he explained.

“And obviously [Milo] is a technology demo that continues to exist, but right now it’s not a game that we’re planning to bring to market.”

Molyneux told VG247 that we’ll be seeing more of Milo and his gal pal Kate next month. From the sound of things, though, we’ll only be seeing – not touching (or flailing in the general direction of, as it were).
Click to expand...

They are probably going to showcase more technologies with Milo.
 
patsu said:
Microsoft has also clarified that Milo is not a game:
Click to expand...

Aaron Greenberg has clarified that it is...

@AaronGreenberg said:
Project Milo absolutely continues in development at Lionhead Studios, it is just not a product we plan to bring to market this holiday.
Click to expand...

@AaronGreenberg said:
The team at Lionhead Studios has always been a center of innovation and will continue to deliver against that charter.
Click to expand...

Plus there are 50 people at Lionhead on the "Team Milo" project...

http://www.joystiq.com/2010/06/29/project-milo-team-about-50-strong-product-not-coming-this-h/

Tommy McClain
 
Well, some of the techs in the original Milo demo has been done in shipping games already (e.g., sketch recognition). Presumably MS can repurpose the collection of techs in Milo into other games. It doesn't have to be "the" Milo title per se.

The "not ready for prime time" ideas will simply be incubated in the labs until they are ready.
 
Yeah, I see Milo as similar to concept cars, created not as a final commercial product but as a backbone to R&D around which ideas can be explored.
 
Yap, that's my hunch too. I remember there's an AI project where the researchers were feeding knowledge to a computer boy (or girl) for years. Milo reminds me of that effort.
 
patsu said:
Well, some of the techs in the original Milo demo has been done in shipping games already (e.g., sketch recognition). Presumably MS can repurpose the collection of techs in Milo into other games. It doesn't have to be "the" Milo title per se.

The "not ready for prime time" ideas will simply be incubated in the labs until they are ready.
Click to expand...

I thought that Milo never did sketch recognition. There was an interview where they were talking about people drawing inappropriate things, and it was pointed out that you never see the image that was purportedly scanned in game you just seen the back of a piece of paper. Not that Kinect couldn't do sketch recognition with the right software. Just more of an example of the smoke and mirrors of the Milo and Kate tech demo.
 
Yes, like I said it's not something Kinect couldn't do, Just that Milo and Kate didn't actually demonstrate it.
 
You must log in or register to reply here.

Similar threads

Arwin
Official E3 Sony Conference Thread 2010
15 16 17
Replies
326
Views
35K
green.pixel
green.pixel
T
Sony PS3 Q+A (Leaked)
11 12 13
Replies
251
Views
70K
one
O
M
Socom: US Navy Seals Fireteam Bravo
Replies
0
Views
3K
mckmas8808
M
A
Full transcript of John Carmack's QuakeCon 2005 Keynote
3 4 5
Replies
98
Views
32K
nemisis
N
Back
Top