Voice Recognition for Mobile Websites

[MacRumors]

Appleinsider points us to an AT&T Research project called Watson. The goal of Watson is to provide speech processing capabilities as a web service.

At AT&T we extended this successful paradigm with speech processing capabilities and created the AT&T WATSON Speech Mashups - a new software framework that casts AT&T’s WATSON speech recognition as a web service to economically bring speech processing technologies to the larger web and mobile developer community.

This web-based speech recognition will work on many mobile multimedia devices including the iPhone.

AT&T provides a video of the YellowPages.com application with speech recognition on an iPhone.

Article Link

[chronzz]

sweet

[snarton]

I hope they use this technology to provide a transcription of voicemail messages, similar to PhoneTag.
--Jeremy

[macduke]

How old is this? They are running one of the first versions of firmware for the iPhone. Look at the old calculator app. Also Safari doesn't have the + icon in the middle. Also...is that a terminal app on there? AT&T does Jailbreaking!! I think thats the biggest story here. Wow.

So this app...just launches a customized version of Safari...or is a plugin for Safari? Wonder how they did that. Maybe Apple gave them special tools for all of this. Why is this just now coming out??

[RoboCop001]

Originally Posted by macduke

How old is this? They are running one of the first versions of firmware for the iPhone. Look at the old calculator app. Also Safari doesn't have the + icon in the middle. Also...is that a terminal app on there? AT&T does Jailbreaking!! I think thats the biggest story here. Wow.

So this app...just launches a customized version of Safari...or is a plugin for Safari? Wonder how they did that. Maybe Apple gave them special tools for all of this. Why is this just now coming out??

Things are different now. They've changed things forever! There's no going back!!!

Where are the other drugs going!?

I... I don't know!! I never knew!! I swear to god!

SWEAR TO ME!

We're tonight's entertainment. Where is Harvey Dent?

Maybe you'd prefer a magic trick. Watch me make this pencil disappear.

[Scooterman1]

Nice demo.
I wonder why they haven't set it to include words like 'TAB' and 'FIND' so you don't have to press ANY buttons. Nice demo, but not very functional if you are driving. If they fully integrate it, it could recognize voice commands, then take you to Google Maps, and then give you 'Turn By Turn'. Now THAT would be an awesome application.

[jasonbuzz]

The swiping looks a little buggy to me....slower than normal and jittery. This is pretty neat, but would be better if you didn't have to tap the screen. That would allow you to find information without looking at the screen so much which of course would be helpful while driving.

[Dmac77]

Why wouldn't I just type the information in, I mean if i have to take the iPhone out of my pocket, then go and tap the field were I want to "voice type"? It would just be faster to type it manually, IMO. Plus I really doubt that it will as stable and accurate as it is in that video.

Don

[aerospace]

My guess is this guy created the program and therefore it is most accurate with his voice, or it took 10 takes for it the program to get all 4 entries right the first time.

VERY cool if it ever worked but voice recognition is still in the dark ages, apple needs to build its own so it works as well as their other creations. Forget touch control, just tell your phone what to do.

"Iphone call David"
"Iphone play Beatles"
"Iphone look up stock price apple"
"Iphone lookup macrumors.com"

Iphone v5 will likely do this lol

[mainstreetmark]

Originally Posted by aerospace

VERY cool if it ever worked but voice recognition is still in the dark ages, apple needs to build its own so it works as well as their other creations. Forget touch control, just tell your phone what to do.

"Iphone call David"
"Iphone play Beatles"
"Iphone look up stock price apple"
"Iphone lookup macrumors.com"

Iphone v5 will likely do this lol

I've seem various people on this forum getting all upset over the lack of "voice dialing", but so far in every voice-recognition software I've ever used, it takes a long time to process and at BEST is wrong 25% of the time. "Call Bart." "*Calling .. Roger...*". Huh? In a quiet room, it never worked well.

Here's an extreme example: http://www.youtube.com/watch?v=KyLqUf4cdwc

I have 10 fingers, and i can use them accurately. Speaking to a computer is still years away.

[crees!]

Originally Posted by aerospace

My guess is this guy created the program and therefore it is most accurate with his voice, or it took 10 takes for it the program to get all 4 entries right the first time.

VERY cool if it ever worked but voice recognition is still in the dark ages, apple needs to build its own so it works as well as their other creations. Forget touch control, just tell your phone what to do.

"Iphone call David"
"Iphone play Beatles"
"Iphone look up stock price apple"
"Iphone lookup macrumors.com"

Iphone v5 will likely do this lol

I believe Apple is working on this. They could have added it just for phone calls like other phones, but think instead when they implement it they'll go all the way like in your examples above; and that's what the hold up is. Doing it right the first time and not just shoving it out the door because customers are use to it.

[chr1s60]

Originally Posted by Dmac77

Why wouldn't I just type the information in, I mean if i have to take the iPhone out of my pocket, then go and tap the field were I want to "voice type"? It would just be faster to type it manually, IMO. Plus I really doubt that it will as stable and accurate as it is in that video.

Don

What if you are driving and need to get the number to somewhere? This would be easier and safer than typing.

I agree that it probably wouldn't be as accurate as the one in the demo.

[gcmexico]

Originally Posted by aerospace

My guess is this guy created the program and therefore it is most accurate with his voice, or it took 10 takes for it the program to get all 4 entries right the first time.

VERY cool if it ever worked but voice recognition is still in the dark ages, apple needs to build its own so it works as well as their other creations. Forget touch control, just tell your phone what to do.

"Iphone call David"
"Iphone play Beatles"
"Iphone look up stock price apple"
"Iphone lookup macrumors.com"

Iphone v5 will likely do this lol

*
but it also has to respond to you with Vicky voice from Imac

me: "Iphone call Betty"
Vicky voice on iphone: "Calling Betty, how is your daying gcmexico?"
me: "Pretty good Vicky, weather is nice"
Vicky voice on iphone: "yes, but based on weather bug it might rain this afternoon, gcmexico connection with Betty established"

now this would be CRAZY!!!

[kdarling]

Originally Posted by macduke

So this app...just launches a customized version of Safari...or is a plugin for Safari? Wonder how they did that. Maybe Apple gave them special tools for all of this. Why is this just now coming out??

There's no special iPhone code. It's just a clever demo, although it could be useful.

It looks like he had already called his Watson test voice server on the phone, then launched his customized Safari webpage over WiFi. The Safari web page talks to the voice server in the background using AJAX.

He clicks one button on the page to send a message to the server to start listening, and clicks again to stop. Then the server returns the text in the background and the webpage displays it.

[aerospace]

Originally Posted by gcmexico

*
but it also has to respond to you with Vicky voice from Imac

me: "Iphone call Betty"
Vicky voice on iphone: "Calling Betty, how is your daying gcmexico?"
me: "Pretty good Vicky, weather is nice"
Vicky voice on iphone: "yes, but based on weather bug it might rain this afternoon, gcmexico connection with Betty established"

now this would be CRAZY!!!

I'd rather my phone not be smart enough to talk to me, haven't you ever seen 2001 space odyssey?

You: "Iphone call Betty"
Hal: "You should really call your mother. Calling mother instead."

You: "iphone launch super mokey ball"
Hal: "Its 8am, time for work, games locked, connecting to boss."

You: "Iphone look up nearest Mcdonalds"
Hal: "You should be on a diet, providing directions to Vegan restaurant."

[brasscat]

The point of voice recognition is that you don't need to use your hands. This guy is tapping all over the place. FAIL

[avigalante]

Originally Posted by brasscat

The point of voice recognition is that you don't need to use your hands. This guy is tapping all over the place. FAIL

I certainly agree.

There's far too much finger input to make this useful.

[geenosr]

Originally Posted by aerospace

I'd rather my phone not be smart enough to talk to me, haven't you ever seen 2001 space odyssey?

You: "Iphone call Betty"
Hal: "You should really call your mother. Calling mother instead."

You: "iphone launch super mokey ball"
Hal: "Its 8am, time for work, games locked, connecting to boss."

You: "Iphone look up nearest Mcdonalds"
Hal: "You should be on a diet, providing directions to Vegan restaurant."

Now that's funny (or scary)

[gcmexico]

Originally Posted by aerospace

I'd rather my phone not be smart enough to talk to me, haven't you ever seen 2001 space odyssey?

You: "Iphone call Betty"
Hal: "You should really call your mother. Calling mother instead."

You: "iphone launch super mokey ball"
Hal: "Its 8am, time for work, games locked, connecting to boss."

You: "Iphone look up nearest Mcdonalds"
Hal: "You should be on a diet, providing directions to Vegan restaurant."

*
pretty funny!

[severe]

Originally Posted by aerospace

I'd rather my phone not be smart enough to talk to me, haven't you ever seen 2001 space odyssey?

You: "Iphone call Betty"
Hal: "You should really call your mother. Calling mother instead."

How bout:

for us single guys...

You: "iPhone call Betty"
Hal: "It's 1 A.M., this isn't gonna look good. You'll regret this in the morning."

I would buy it right now.

[nickXedge]

Originally Posted by macduke

How old is this? They are running one of the first versions of firmware for the iPhone. Look at the old calculator app. Also Safari doesn't have the + icon in the middle. Also...is that a terminal app on there? AT&T does Jailbreaking!! I think thats the biggest story here. Wow.

So this app...just launches a customized version of Safari...or is a plugin for Safari? Wonder how they did that. Maybe Apple gave them special tools for all of this. Why is this just now coming out??

Despite the time that this article is written on MacRumors, it should be clear to see that this is not an iPhone Application. Pointed out by you, the older calculator, the older version of Safari. Notice there is no App Store button on the home screen.

It should be obvious to see that this is not a third party YellowPages application, but instead this is a WebApp for iPhone. So no, AT&T does not Jailbreak, and they do not make applications for YellowPages or ones that "launch customized versions of Safari."

[cubedweller]

Originally Posted by nickXedge

So no, AT&T does not Jailbreak, and they do not make applications for YellowPages or ones that "launch customized versions of Safari."

What about that little Terminal icon?

[kdarling]

Originally Posted by nickXedge

It should be obvious to see that this is not a third party YellowPages application, but instead this is a WebApp for iPhone.

Yes, as I pointed out above in this thread, it looks like a web page tied to a voice server that the guy called on the phone.

So no, AT&T does not Jailbreak, ....

It seems pretty clear that a programmer at AT&T used an iPhone that he had jailbroken. He might be catching some minor ribbing right now about that being visible.

[Aznsensation45]

wow. at&t doing something cool? that's unheard of!

[AJSchmidt]

Nothing New at all. Yahoo has had a voice search for BlackBerrys for a year now...

http://mobile.yahoo.com/onesearch/voice