AT&T is not currently planning to use this tech to manipulate current iPhone apps (Contacts? Maps? Mail?) but instead will deploy it in web services for a number of net-enabled handsets. This is a somewhat curious choice for AT&T, but it would be difficult to implement system-wide speech recognition without either modifying existing apps or running a (currently disallowed) background service to catch commands. Speech Mashups will be an interesting service for the other handsets it shows up on, but they already have simple voice commands. By building secondary voice capabilities like this for a phone without basic ones, AT&T has inadvertently highlighted one of Apple's most irritating restrictions on iPhone development. [Gadget Lab]

Pretty impressive, and saves all that fiddling around with buttons when you should be busy controlling your car. It's got background noise suppression, so apparently it can cope with driving noise. And its independent battery runs it for 10 hours, without affecting the iPod.

There's just one flaw: its price. At $159 it's more than a 4GB nano itself, and that seems a little crazy. They are planning on releasing new versions for other iPods and the iPhone "as soon as possible," but this one will be out in December. [DirectVoxx]

Five Big Wins

The internet - You know, ARPANET, or a communications network that heals thyself. The whole porn-music-movie triangle trade was not in the original blueprint.

GPS - The idea that satellites up there could tell us where we are down here is as old as Sputnik, and DARPA had an attempt or two before getting it right with GPS. Again, not originally purposed for civilians and their "points of interest"—more about targeting ballistic warheads, but you know, potato, po-tah-to.

Speech translation - Soldiers in Iraq use handheld machine translators to aid in communication with some degree of success, meaning it's only a matter of time before some German tourists ask you to speak into the microphone.

Stealth planes - Stealth airplane technology was so sneaky it even snuck up on Air Force top brass, who were apparently shocked to learn about the prototype for the original F-117.

Gallium arsenide - Yep, some DARPA egghead discovered you can do more with arsenic than poison KGB triple-agents. It's now found in all kinds of everyday electronics. That may not last, though, because environmentally speaking it's still some pretty nasty stuff.

Five That Didn't Quite Make It

Hafnium bombs - DARPA put $7 million into researching a bomb with massive initial devastation but no radioactive fallout, but alas, it didn't ever work. Apparently if you want the good, you gotta take the bad.

The mechanical elephant - Hannibal would have been proud of this one: During the Vietnam War, some dudes at DARPA wanted to take terrain-friendly robot elephants into the jungle. Even DARPA's director was embarrassed.

Telepathic spies - People who claimed psychic powers were on the receiving end of a lot of government funding in the 1970s. Even though the project was a failure, getting rich by pretending to be psychic does seem to suggest a kind of sixth sense.

FutureMap - Apparently a bunch of Dick Cheney's friends betting on terror targets was considered grotesque by some people. I've still got $30K riding on Dubuque.

Project Orion - This is a nerd favorite: it's the spaceship powered by atomic-bomb turds. I think everyone was sad to see that one go.

Check out the New Scientist story for the real deal behind these 10 projects, and a look at some crazy projects that might make the list in the next 50 years. [New Scientist]

When you sign up for PhoneTag, which formerly went by the incredibly nerdy name SimulScribe, you get a phone number to add to your GrandCentral account. When you miss a call, PhoneTag picks it up and records the message. The speech is converted to text, which is then sent to you via email or text message, along with an audio attachment of the voicemail—two things Grand Central doesn't do. If you add your contact list to your PhoneTag account, the messages will appear from the senders so you can reply through email right away. PhoneTag is free for a month, and plans range from $0.35/message to $30/month for unlimited messages. [PhoneTag]

"Your car could know a heck of a lot about you," says Clifford Nass, things like where you go, what speed you drive, what music you listen to. And, of course, companies will pay for that info, but Nass is also wondering if your car could speak back to you, would it result in you driving more safely, or even cheer you up if you were depressed?

Nass' research is interesting to insurers, of course, because if your car could detect that you consistently drove too fast or were often angrily swearing at the traffic while you wrenched at the wheel, they might be tempted to up your premiums. But Nass also sees our future cars being like dumber versions of KITT, behaving more like a co-pilot and actually speaking back to us. Perhaps pointing out that there's a pedestrian in the road ahead, or warning you that you're driving too aggressively.

Some of Nass' results are fascinating:
•Depressed drivers drive better when their car reacts as if it too is depressed
•Cars that speak in complex sentences tend to make you pay attention, rather than zoning out
•Older drivers prefer younger voices
•The car's voice is important: BMW had to recall Series 5 cars because male German drivers didn't listen to its female voice
•Drivers feel closer to the computer if they believe it's embedded in the car, rather than a remote device.

That last one seems like an extension of our personification of cars: I've named every one of my cars, and I bet you guys do too. And while a talking, monitoring, advice-giving car may be some people's idea of a nightmare, you can see that the idea has potential for improving safety: "You can imagine how much it costs for anti-lock brakes. But what if you could simply change the voice in the car?" points out Nass. What do you think, guys? Talking cars FTW? ... or KITT-like cars as nightmare? [Physorg]

For the brain to send the right nerve impulses, it apparently requires the user to think about speaking in a particular way—one designer calls it "a level above thinking." Despite the learning curve this causes, requiring users to go through lots of training, it brings another benefit: you can intersperse talking normally with voiceless speech.

The company has been working on the technology for a while, starting by using it to control an electric wheelchair. For the time being, though, the speech system is a technology in its infancy. The processing delay is evident from the video, and it has a limited vocabulary of only 150 words and phrases. By the end of 2008, the company is hoping to release a new version that recognizes the phonemes that comprise normal words, effectively giving it an unlimited vocabulary. [New Scientist]

The Catch: It only works with books made specifically for the device, which are put together by Aigo themselves. Last time we checked, Aigo wasn't big in the publishing industry, so good try guys. Perhaps we shouldn't knock it. We are sure it will work great with tourist paraphernalia and the like, (they had an Olympics guide on show, which kind of makes sense). No word on pricing or shipping dates at present. [Aigo]

Although it will offer service, ThePudding isn't trying to claim a piece of the pie that Skype, Vonage and the cable companies have been wrassling over for years. According to the AP, it hopes to "license its speech-recognition service to other companies that use Voice over Internet Protocol." But AP tech writer Peter Svensson had mixed results when testing ThePudding's speech recognition:

"Relevant ads appeared when this reporter talked about restaurants and computers, but the software was oddly insistent that he should seek a career as a social worker, showing multiple ads and links pointing to that field."

It's for real (see update below). Sure will be nice to be able to use speech commands with the iPhone, telling it to call so-and-so on those mic-equipped earbuds while keeping the phone in the pocket.

UPDATE: We got this exclusive info from Chris LeBlanc at VoiceSignal: "It works just like our other apps, so it's speaker independent and needs no training at all to recognize names, numbers and general speech. We will demo the continuous voicemode on the iPhone soon, I've already seen it — and it's nearly ready." Chris has promised us a demo copy, so we'll give you a first look as soon as it arrives. [VoiceSignal]

Besides being a source of comic relief, calling out the street names with this text-to-speech capability can actually be useful—keeping your eyes on the road and away from that map on the Nuvi 260's smallish 3.5-inch screen. Garmin says this Nuvi 260 will be shipping this month, and it'll retail for just under $500. And that's a "budget" unit? Expect (hope for) discounts forthwith. [Press Release]

The efforts of the team are coming to fruition now, following years of research in the field. They have been able to integrate the multiple speech recognition with enhanced existing technologies. This has allowed robots to recognise the direction and hence categorise speech input qualitatively according to each individual.

The team has already incorporated the developments into the much-loved Honda Asimo and another robot called Robovie R2, seen in the video.

The practical applications are staggering, but surly the most pertinent is being able to finally have a well-communicated, meaningful threesome with your two sex doll robots. You sicko. Hit the video to see it in action. (The technology, not the threesome. You sicko). [Technabob]

Our first test call was from a landscaping service that we used here at the Midwest Test Facility three years ago. See if you can decipher the meaning of this message:

"Good Afternoon this is linda for me well branded sign I'm i'm getting a hold of Truly whites residents or Company Evil goes out your neighborhood and he said that one you're properties looks like it could Use some pruning it's been three years now since we were out there And well wondering if you'd wanna a proposal from us Please give me a call At two six two two four four Nine four zero zero And let us know if that Is something you would like for us to do Thanks I'll wait for your call back up right of"

Then a second message came in, and this one fared a little better:

"Hey, it's Kim. I'm calling about lunch today. I was just calling to see if it was alright if we met at 12:30 instead of 12, 'cause I (??) have to work till 12 and I'm (??) but I can be there by 12:30, no problem. If that gonna be a problem for you, can you please give me a call back. Otherwise, I will see you at the (??)'s place at 12:30. I look forward to seeing you. I hope that's gonna be okay. Talk to you soon. Bye-bye."

We wish the implementation was a little closer to perfect for these transcriptions, but the idea of having your voicemails delivered to you in text form is highly appealing. Imagine in a meeting, you could a quickly glance at the text of all your voicemails and immediately catch up with what's going on.

However, that $.25 price for each transcription is just not cost-effective enough. If you get a dozen voicemails a day, your monthly tab would hover around $90 for this convenience. We're thinking more along the lines of 5 cents apiece would make it more practical. We're canceling the service right away.

Language acquisition has long been seen as a specialty of the brain—a strength of the species, if you will. To re-create one of our most distinct qualities in a lab setting is either impressive for computers or pitiful for humans. Then again, be comforted knowing that when computers do speak, they'll sound just like Arnold Schwarzenegger, anyway. [arstechnica]

Its maker says it's ruggedized, unphased by a fumble onto pavement or a torrential rainstorm, and can translate phrases you speak in English into any language, and then translate back into English whatever people say to you. Hmm, that's a lot of languages—perhaps that's an exaggeration. Does it work, and how expensive is this thing?

Just like the latest Version 9 of Dragon NaturallySpeaking, you don't have to train it to understand your speech, and any male or female can use it right out of the box. It was developed for military use, so it must have some high technology inside, but that gives you a clue why the thing is so expensive: $2000.

Even so, it looks like it would be fun to take this baby out for a conversation or two. But then, there's no word whether it'll be able to translate GuySpeak into WomanSpeak. – Charlie White

Translator Gadget: Phraselator P2 by VoxTec [Product Reviews Net]

The news came straight from Logitech's reps, who confirmed that their next Harmonies will use IBM's ViaVoice recognition software so that you'll be able to change channels by barking out commands. It gets better though.

The remotes will also rely on fingerprint readers to load customized preferences for every person in your household. Lastly, Logitech is working on a search function that'll let you load up songs from your media library by simply saying the name of the band you want to hear. So in other words, "play Sinatra" would launch your Sinatra tunes. I tend to stay away from universal remotes 'cause of their price, but something like this might be worth the splurge. – Louis Ramirez

Harmony Remotes to Include Speech Recognition, Search [PC Mag via Gadget Lab]

This is actually a pretty cool service, seeing as other transcription services we've seen are either expensive or strange to use. SpinVox has lined up Cincinnati Bell and Skype, and are working on some deals with major carriers now (no details yet). We know many people who don't bother listening to voicemails because it requires dialing in, pressing buttons, and listening—these are very lazy people.

Other cool SpinVox usages are sending memos and broadcast messages from your phone by calling a number and speaking. Sounds like a great way to send messages to your husband to pick up some tampons. – Jason Chen

Product Page [Spinvox]

Microsoft [Thanks Eric!]

All that coolness doesn't come cheap: it'll cost you a hungry grocery shopping trip's worth of moola to buy it ($150). – Charlie White

SmartShopper: Tell It What You Want, It Prints a List [Coolness Roundup]

This waistband speech amplifier makes you heard for up to 50 meters! In America, we call meters yards, if I understand it correctly! That makes the range half the length of a football field, the field that America's game, football, is played on! Or wait, maybe baseball is America's game! In any case, football is quite American as well!

Let everyone know just how you feel! Only $125 American dollars! –Adam Frucci

Product Page [via Uber-Review]

If this feature indeed works the way they describe, this could be the best phone ever. If not, it's just another phone with a 2-megapixel camera that's available only in China. – Jason Chen

PhoneDaily [via Mobile Mentalism]

The Wordy Birdy Digital Speech Trainer makes that training process a lot easier by repeating, at intervals between 30 seconds and 20 minutes, whatever word or phrase you've recorded into it. Leave it on until your poor bird (and you) have had enough. Just be sure it's not within earshot of the neighbors—it'll drive them nuts. It's $22. – Charlie White

Product page [That Pet Place, via OhGizmo]

The service will cost as much as a standard SMS and will probably be used once by every Sprint subscriber and quickly forgotten.

Sprint Unveils Text Messaging to Landline Service [LocalTechWire via Textually]

Epson Chip Can Read Text [I4U]

The mobile phone that slows down speech [T3]