AT&T developing voice-controlled iPhone apps (video)

[AppleInsider]

AT&T has developed a software trick that will let modern mobile handsets, including Apple Inc's iPhone, recognize voice commands without the need for specialized voice recognition software.

The research project is based on a new version of AT&T's WATSON speech recognition engine, dubbed Speech Mashups, that puts the entire feature on the web as a service that can be called upon from anywhere a high-speed Internet connection is possible.

As long as the software used to access Speech Mashups obeys certain web standards, particularly an AJAX framework and JavaScript, the technology can capture voice commands, interpret them at a remote server, and send them back to the device in a language a website or program can understand -- all without installing a dedicated app or plugin.

The telecoms company says the technology can be used for IP-based TV boxes as well as BlackBerries and smartphones, but draws most of its focus to the iPhone -- a device which (unlike the BlackBerry) has no native voice recognition of its own and, until the release of iPhone 2.0 firmware, had no support for the feature even through isolated native apps.

In a prototype mobile version of the YellowPages website, AT&T in a research video shows an iPhone user entering the business name and location into text fields on the page just by speaking them at the appropriate times. *While typing would work in such a case, the company claims that voicing the information is faster and more convenient -- especially when driving.



This solution is limited and excludes iPhones without a sufficiently fast connection to AT&T's servers or for native applications that don't include web code; many of Apple's own applications, for example, wouldn't function with the feature. *As-is, the technology doesn't satisfy frequent requests for voice dialing or other direct speech recognition features.

Still, while the development is limited in scope and remains in AT&T's labs, the development potentially opens up both web apps and some native iPhone apps to a feature that even Apple itself has yet to program into its own devices.

[ View this article at AppleInsider.com ]

[iPhone91]

It might be really cool, if they make it work well and across the whole iPhone. Still, the better solution would be for Apple to just write in Voice Commands into the next software release. Add MMS and Video Recording, then (almost) all the critics will be silenced!

Steve

[AHeneen]

Completely irrelevant, but interesting...I was very confused as to how this showed up at "12:00am EST," when my computer displays 11:27PM (Eastern). Not only that, but we're now on Eastern DAYLIGHT Time which is EST+1...so actual EST is 10:27 PM. What's wrong AI, these Panasonic ToughBook, Kodak printer, and T-mobile ads not garnering enough clicks to afford decent software?

[Slewis]

Null.

[iVlad]

Aggg...why ATT. Apple can do it twice as better. Its already done in OS X. If only Apple had some more time and engineers the whole iPhone would listen to you.


Wait....Sounds like a new feature for future update.=) lol

[SpamSandwich]

Might be nice if at&t opened this up as an online Mac app also, perhaps a widget?

[winterspan]

Voice Dialing really isn't difficult, I'm actually very surprised if there is not a popular program in the app store for that? Is this forbidden in the app store or something??? If not, I guarantee one is in development..

[xc3ll]

This isn't really relevant, but it looks like the guy has a terminal on his iPhone. It seems doubtful though that an official video (From AT&T nonetheless) would feature a jailbroken iPhone. Can anyone shed some light on this?

Is it one of those web terminals I've heard about?

[Ireland]

Originally Posted by iPhone91

It might be really cool, if they make it work well and across the whole iPhone. Still, the better solution would be for Apple to just write in Voice Commands into the next software release. Add MMS and Video Recording, then (almost) all the critics will be silenced!

Steve

Except the critics who want a good camera in their phone.

[Ireland]

Originally Posted by xc3ll

This isn't really relevant, but it looks like the guy has a terminal on his iPhone. It seems doubtful though that an official video (From AT&T nonetheless) would feature a jailbroken iPhone. Can anyone shed some light on this?

Is it one of those web terminals I've heard about?

I can. The video is from AT&T. That's all you need to know.

[nvidia2008]

I want the voice activated feature where I can say, "Unlock my F*KING iPHONE 3G you BASTARDS"!!!

[Ireland]

Originally Posted by nvidia2008

I want the voice activated feature where I can say, "Unlock my F*KING iPHONE 3G you BASTARDS"!!!

iPwnageTool, but you have to click while you talk.

[NOFEER]

Originally Posted by nvidia2008

I want the voice activated feature where I can say, "Unlock my F*KING iPHONE 3G you BASTARDS"!!!

its about time someone said just that
voice dialing, how about fingerprint reading so you don't have to swipe then put in you pin then talk, come on, the ui needs some buff.
yea
voice dialing, commands, finger print reader app--now only $200

[Iphtashu Fitz]

AT&T is playing catchup... A company called Vlingo has been doing this sort of thing for a while. The company has a bunch of speech scientists from MIT whose previous speech rec company was acquired by Nuance. They've already got a full speech app available for BlackBerry's (you can download it from their home page) and I've heard rumors that they have an early iPhone version being tested. In a nutshell, if you want to send an e-mail, reply to an e-mail, etc. just press & hold the "send" button and start talking. It streams your audio to a server that converts the speech to text and returns the text back to the e-mail application (or whichever app you happen to be running at the time). It's pretty slick stuff. I've seen it demoed a few times. If you've got a BlackBerry I'd suggest downloading the Vlingo app and trying it out.

[Iphtashu Fitz]

Originally Posted by iPhone91

It might be really cool, if they make it work well and across the whole iPhone. Still, the better solution would be for Apple to just write in Voice Commands into the next software release. Add MMS and Video Recording, then (almost) all the critics will be silenced!

This has the potential of a LOT more than just voice commands. Imagine the ability to dictate entire e-mails by voice rather than typing. See my previous post about the company Vlingo that's already doing this. Which would you rather do, type in a 200 character e-mail or simply speak it to your phone and have it translated directly into text? I've seen the Vlingo stuff in action on a BlackBerry and even in a rather noisy public environment it can be remarkably accurate.

[Techslacker]

Originally Posted by Iphtashu Fitz

This has the potential of a LOT more than just voice commands. Imagine the ability to dictate entire e-mails by voice rather than typing. See my previous post about the company Vlingo that's already doing this. Which would you rather do, type in a 200 character e-mail or simply speak it to your phone and have it translated directly into text? I've seen the Vlingo stuff in action on a BlackBerry and even in a rather noisy public environment it can be remarkably accurate.

So we've got instant messaging and texting that gets used many times where email could have worked and now people want to use email with voice recognition where voicemail could work. Most people struggle with what apps to use for what function...nothing like confusing them more.

And, yeah, I understand the reasoning some want this....just being a bit sarcastic about it. It all sounds as crazy as using a web browser to write a paper.

[dagamer34]

Anyone notice that that iPhone is running an extremely old firmware (1.0-1.02, check the calculator icon) and it's been jailbroken (where'd they get Terminal from)?

[Adam Venier]

So how did AT&T manage to record audio from within HTML?

This is easy to do from applications created with the SDK but I didn't think it was possible from within Safari. Thanks.

[NanoAkron]

Originally Posted by iPhone91

It might be really cool, if they make it work well and across the whole iPhone. Still, the better solution would be for Apple to just write in Voice Commands into the next software release. Add MMS and Video Recording, then (almost) all the critics will be silenced!

Steve

And the ability to forward contacts as address cards/business cards via texts/mms/bluetooth. A feature that is greatly missed.

How about proper editing features for emails so that I don't have to forward an entire conversation, just the attachment?

[bloggerblog]

The whole purpose of voice-dialing, for me at-least, is so I don't have to click-around on my iPhone while I'm driving. Maybe if they did a double-click-and-hold on on the lanyard then speak the name or the actual number, and after releasing the button it will search the contact. It will then verify the contact by speaking it back, and with a single click allows you to accept or a repeated double-click-and-hold allows you to repeat.

They should implement something similar to that.

[macologist]

And if one calls with a foreign accent, they'll have a nervous breakdown! Never mind someone with a really thick accent!

Can I have a Sadim wisa woenuts and diet coke to go?!

Sadim wisa woenuts = Shrimp with walnuts...

Flied Lice anyone? Cheesburgah, Cheesburgah!

Oops, back to work!

[JeffDM]

Originally Posted by Ireland

Except the critics who want a good camera in their phone.

They might be able to get something with better specs, but I doubt they'll ever get a good camera in their phone any time soon.

[JeffDM]

Originally Posted by macologist

And if one calls with a foreign accent, they'll have a nervous breakdown! Never mind someone with a really thick accent!

Can I have a Sadim wisa woenuts and diet coke to go?!

Sadim wisa woenuts = Shrimp with walnuts...

Flied Lice anyone? Cheesburgah, Cheesburgah!

Oops, back to work!

I get the rest, but how does shrimp become sadim?

Accents will probably always be a problem, which is probably part of why speech recognition needs to be trained to the user.

[8CoreWhore]

It'll be ready in 5 years and they'll charge us for it.

[NOFEER]

Originally Posted by macologist

And if one calls with a foreign accent, they'll have a nervous breakdown! Never mind someone with a really thick accent!

Can I have a Sadim wisa woenuts and diet coke to go?!

Sadim wisa woenuts = Shrimp with walnuts...

Flied Lice anyone? Cheesburgah, Cheesburgah!

Oops, back to work!

there isnt a problem when the voice dialing is on the phone like my ancient v551, you record and tag the voice, that's why voice recognition is such a pain ( i work with it hourly, its so bad that we have an editor to make the corrections)