Log in

View Full Version : Talk to the Machine


Jason Dunn
11-14-2002, 10:44 PM
<a href="http://www.spectrum.ieee.org/WEBONLY/publicfeature/sep02/voic.html">http://www.spectrum.ieee.org/WEBONLY/publicfeature/sep02/voic.html</a><br /><br />If you've ever wanted to know exactly how voice recognition works, give this article a read - it's a very detailed write-up of the current status of voice recognition, and how the technology works. Very enlightening!<br /><br />"Max Huang says he has something cool to show me. I'm skeptical: he's holding in his hand what looks like a PDA. It is a PDA, a Compaq 3600, to be exact, unadorned and, to my eyes, unremarkable. What's special is what's inside: this PDA understands what you say. <br /><br />Huang and his colleagues at the Philips Speech Processing office in Taipei, Taiwan, have streamlined the company's standard speech recognition engine, meant for servers and PCs, to run instead on a PDA. It's just a prototype, Huang says, but the Mandarin-language recognizer can distinguish about 40 000 words and still not tax the Compaq's memory, power, or processing. With it, Huang can access his address book, schedule appointments, and dictate e-mail. Considering the alternative—poking away at the device's tiny display with a skinny stylus—I'm starting to be convinced: this does seem pretty cool."

mookie123
11-14-2002, 11:00 PM
awesome, now all I need is learning chinese so any iPAQ will understand me. mmm...

ThomasC22
11-14-2002, 11:02 PM
hmmm...the promise of voice recognition on a PDA...haven't heard that before :roll:

Jeff Rutledge
11-15-2002, 01:30 AM
I think it's an inevitable evolution, but unfortunately I think it's still at least 3 years away. Hopefully I'm being pessimistic, but I think that's how long we have to wait still.

ECOslin
11-15-2002, 02:13 AM
I'm not so worried about voice recognition yet on my PDA. I would like a text to speech version on the version of Microsoft Reader, or any reader, for my PDA. Something with a human voice and human parsing.

Although asking, by voice, my PDA what the weather is outside, and having it answer, by voice, would be neat.

Edward

mscdex
11-15-2002, 04:34 AM
well, the Pocket ViaVoice has been out for quite some time now, I tried it once, and it was ok, i guess.

ThomasC22
11-15-2002, 05:16 AM
well, the Pocket ViaVoice has been out for quite some time now, I tried it once, and it was ok, i guess.

If by ok you mean "understands about 10% of what you say" then yes.

Jeff Rutledge
11-15-2002, 05:40 AM
well, the Pocket ViaVoice has been out for quite some time now, I tried it once, and it was ok, i guess.

If by ok you mean "understands about 10% of what you say" then yes.

10%???? Wow! What did you do to teach it so well?!?!? :lol:

mscdex
11-15-2002, 05:53 AM
well, the Pocket ViaVoice has been out for quite some time now, I tried it once, and it was ok, i guess.

If by ok you mean "understands about 10% of what you say" then yes.

Well, It was about 50% of what I said or so. I can't quite remember since I had installed it a long while ago, and then I had uninstalled it because I guess I'm just a stylus person. And if my memory serves me well, another flaw pocket viavoice had was that it was limited to certain programs.

ThomasC22
11-15-2002, 05:59 AM
Well, It was about 50% of what I said or so. I can't quite remember since I had installed it a long while ago, and then I had uninstalled it because I guess I'm just a stylus person. And if my memory serves me well, another flaw pocket viavoice had was that it was limited to certain programs.

aaaaaannnnnnddddd iiiiiiittttttt mmmmmmaaaaaaaaddddddddeeeee yyyyyyoooouuuurrrr ppppppoooccckkkeeeetttpppppcccc vvvvveeeeeerrrrryyyyy sssssslllllloooooowwww!!!!!!

Certified Optimist
11-15-2002, 09:37 AM
At least with a Pocket PC you can see that someone is speaking to something...

With those hands-free kits, you're not sure if it's just another nutter or someone actually carrying out a normal conversation. Especially went it's night...

jizmo
11-15-2002, 10:05 AM
Dictating e-mails, running programs with speech .. yup, all "nice" features.

But where speech recognition would really become handy, is translating. Just think about it: you say a word, PDA understands it, looks it up in dictionary and speaks it out trough speech-syntesis.

This way you could communicate understandably almost any language in the world. Not exactly a babelfish in your ear, but a killer application for sure.

/jizmo

ECOslin
11-15-2002, 11:35 AM
Gee Jizmo, at least you'd know why the Gendarme suddenly grabbed you to take you away.

'Aucun Aucun Monsieur, mon ordinateur de poche a dit que j'étais un espion','Je ne suis pas un espion,Il couche! ,Il couche!

Edward

(No No Mister, my pocket computer said that I was a spy, I am not a spy, It lies! It lies!)

mscdex
11-15-2002, 12:34 PM
Well, It was about 50% of what I said or so. I can't quite remember since I had installed it a long while ago, and then I had uninstalled it because I guess I'm just a stylus person. And if my memory serves me well, another flaw pocket viavoice had was that it was limited to certain programs.

aaaaaannnnnnddddd iiiiiiittttttt mmmmmmaaaaaaaaddddddddeeeee yyyyyyoooouuuurrrr ppppppoooccckkkeeeetttpppppcccc vvvvveeeeeerrrrryyyyy sssssslllllloooooowwww!!!!!!

iiiiiiiiiiiiiiiiiiii ddddddddddddooooooooooooonnnnnnnnnnnnn''''''''''''''''ttttttttttttttttttt rrrrrrrrrreeeeeeeeeemmmmmmmmmeeeeeeeemmmmmmbbbbbbbbbeeeeeeeeeeeerrrrrrrrrrrrr

Sven Johannsen
11-15-2002, 04:00 PM
Can't believe no one has commented on the picture that accompanies the article.

Jeff Rutledge
11-15-2002, 04:25 PM
Can't believe no one has commented on the picture that accompanies the article.

Maybe that's the key. You get 100% recognition if you whisper to your PDA like a school girl telling a secret.
:wink:

daveshih
11-15-2002, 04:35 PM
It does sound cool, especially considering it's not recognizing English (which has seen heavy voice-recognition development for a long time) but mandarin Chinese, which has more words than you can imagine. To a native Chinese speaker, like me, who have searched high and low for a suitable voice-recognition system on his/her pc, this is great news.

BTW, kudos to Microsoft for publishing a Chinese voice-recognition engine for free. Even though it's still rough, but at least workable.

Still, on a Pocket PC, this is no small accomplishment. I especially like the part where it integrates directly into pocket outlook's appointment scheme of things. To me, this solves 80% of what other voice-recognition system has tried to do and failed.

For example,
"Now I will encapsulate everything into one sentence," Huang says. " 'I want to see a movie with Lifen Yeh at 10 o'clock tomorrow.' " Within a second or two, the form has been filled out, except for the "where" field. "I didn't say where the movie is, so it left that blank," he explains. To confirm the appointment, Huang opens the e-mail program by simply saying "Send it to her." The recognizer understands that "her" means Lifen Yeh

That along, I think, is a great achievement, and will greatly enhance my Pocket PC's usefulness.

David

ECOslin
11-15-2002, 04:50 PM
I've found that where I would want to use voice recognition, is usually not the environment to use voice recognition.

Open market with hundreds bargaining one person louder than the next.
Outside a Hotel/Museum or whatever with loud diesel street traffic noise.
Train station with PA blasting in 2 or 3 languages.
Some open restaurants during lunch hour.
Airconditioning whine and electronics buzz(phone and networking closet).
Some aircraft I've been on you might have well ridden the engines for all the sound insulation that was used.

When you get time and space to think, there would be someone yacking at a cell phone.

I'd probably rather use pictures, to ask someone for something, like that commercial.

Edward
(Commercial where some dude is in an Asian market trying to use hand signals to find a toilet, until some lady emails a photo of toilet to his graphics enabled cell phone.)

ThomasC22
11-15-2002, 10:44 PM
Can't believe no one has commented on the picture that accompanies the article.

he's saying "you suck, you suuuuck!"

Seriously, I'm wondering if the quality of this software would improve on a PPC (being if the picture is accurate it currently runs on a 33mhz processor)

Merlion
11-16-2002, 02:12 AM
MS was touting voice recognition for the WinCE PDAs (PalmPCs/Palm-size PCs) even before the 1st devices hit the market. Many, if not all the original machines, like the E-10, came with a voice-recognition program. I think it was powered by Dragon-Speaking System(?). Somehow, it didn't work that well & it slowly fizzled out.

Voice-recognition (& hand-writing recognition) for Chinese is great coz it's very troublesome & exasperating to peck on the keyboard multiple times just to build up one single Chinese character, since Chinese is not a phonetic language. Even more so, I think Chinese, & some other east Asian langs, & other pidgin & creole languages (& also some International Auxiliary Languages like Glosa), have an extremely simple grammatical structure (to the extend that many of their native speakers consider them as being "grammerless"). There're no inflections, no tenses, etc. So, I'm thinking that concept-based languages like Chinese would lend itself very well to voice recognition, coz the program wouldn't have to handle the extremely complicated grammatical issues of ethnic European langs, especially with the inconsistant constructs & special cases, & the compounded complexities with the infinite combinations of all those rules.

This is regarding voice recognition for speech of course, & not 'voice command' recognition where the context & grammer are non-issues.

Hmmm .. then again, not being a phonetic language, many many Chinese characters actually have the same pronounciation. It's the writing of the individual characters that gives them meanings, not the combination of a system of phonetical alphabets that give them meanings, so the spoken form of each character is very much more ambiguous then the written form. So aurally, the program will have to rely on the combination of characters & the context to convey meanings. This makes it harder. :?