Slashdot Mirror
Google Voice Search May be Coming Soon
vitaly.friedman writes "The master of text-based search could be looking to lend a voice to Internet users everywhere, or so it appears based on Google's latest patent. Patent #7,027,987, issued today by the US Patent and Trademark Office, covers a 'Voice interface for a search engine.'"
lend a lend a voice?
Do Slashdot editors submit stories with voice recognition software?
i) Take two existing technologies, such as voice recognition and search engines.
ii) Put the two together and patent the result.
iii) Profit!
I'd say that if Google had any serious plans to do anything with this, then we would have heard something about it already (something more than a paper written 4 years ago). Although the patent was only just issued today, it was filed over 5 years ago.
It seems like a pretty big logical leap to say that it is "coming soon" based on the fact that the USPTO finally got their butts into gear.
It's probably just an idea they though was cool at the time, and should be patented in case they want to use it some time.
I doubt speaking the word "sex" to my pc at work will get me a raise anytime soon... I'll stick with traditional methods of finding my porn... =)
It can be interpreted either way. Either a search engine for audio files with speech recognition, which could be used to index podcasts and news streams; or a voice-driven interface instead of a visual one. Now which is it?
i ate crayons when i was a kid and now i have two braincells and the blue ones taste nicer
Haven't they already done a thing like this before? Maybe they've only just patented it. Either way I don't see a story here.
Guy asked me for a quarter for a cup of coffee. So I bit him.
Economic terrorism, that is.
Do they mean that they patented using voice recognition software to work with the Search textbox? Is this some sort of technology to take voice waveforms into the search engine and form audio search patterns? Is this about taking a media clip and finding the source media?
What? It's so vague.
From May 2002:
http://labs1.google.com/gvs.html
Google voice has been and gone from the Labs pages; make a phonecall, get some results. I think from that it's a no-brainer that it's coming eventually... did we really need a patent as confirmation?
Especially from the image search page.
You have been warned.
liqbase
"'Voice interface for a search engine.'""
Great! You get to curse at your search engine.
Great! Now the government will be able to not only track my search results...but also attach my voice to it! Not that I'm scared or anything...
I remember that Google had a link that you could dial in and give it a search term. You would then go to their "voice results" page and see what it turned up. I assume that the patent is based on this? I haven't looked at it in a few years and I don't even know if the page is still there, but this really isn't anything new, just that the patent was now granted.
In other news:
Patent #193983891 prevents competitors from using fonts to display text on a search engine
If they don't patent it, somebody else will!
-:sigma.SB
WARN
THERE IS ANOTHER SYSTEM
In the library...
.....porn ...porn ...Search item confirmed. Commencing search for "porn".
Google Voice: Please state what it is that you would like to search for.
MAN (whispering):
GV: I'm sorry, I did not understand you. Please state your search item again.
MAN (whispering):
GV: I'm sorry, I did not understand you. Please state your search item again.
MAN (whispering): porn
GV: I'm sorry, I did not understand you. Please state your search item again.
MAN (shouting): PORN! I want to search for PORN you stupid ass computer!
(stares from every person in the vicinity)
GV:
Just think of all the entertaining stories that wide-spread voice recognition will bring us ^_~
Hero of Allacrost, a FOSS RPG for *NIX/*BSD/OS X/Win
My first read (after reading the ARTICLE :P) is that this isn't voice recognition - its the old "wisdom of croweds" thing.
I always found Google's "Did you mean ____?" to be better than any spell checker (pretty sure its a distance metric thing based on LOTS of mistyped input and follow-up for real users) - don't see a reason why that couldn't apply to voice...
(non-trivially, probably, but still)
--
graphicallyspeaking
graphically speaking
Perhaps its the old "wisdom of croweds" thing, ala Google's "Did you mean ____?"
That almost always crushes standard spell checker results (pretty sure its heavily augmented with some kind of distance metric thing based on LOTS of mistyped input and follow-up for real users) - don't see a reason why that couldn't apply to voice search...
(non-trivially, probably, but still)
--
graphicallyspeaking
graphically speaking
Is it called Opera?
Opera has built in voice recognition already, so could someone clairify the point of building this into a search engine?
"My Voice is My Passport, Verify Me..."
Wow, yet another garbage patent that does nothing except combine two existing things (search engines and voice-controlled software). Is there anyone in the world who seriously believes that without giving Google a 20-year monopoly on this, nobody would've thought of this?
quidquid latine dictum sit altum videtur.
...they'd better have tea on this stinkin' planet.
That is when this was applied for. Since we haven't seen it yet, I would not put a lot of hope (or concern?) into this showing up anytime soon.
"Some days you just can't get rid of a bomb."
I know of 2 people that have been searching the internet from voic command for over 6 years now.
Is this prior art ?
Google is OBVIOUSLY not the first to do it or even think of it.
I can show you a half dozen Sc-Fi episode that have touched on this as well.
How can you patent a communication medium's use ????
I can't wait to see his face when google toolbar will start throwing pages at him.
hilarious
gives me a raise just thinking about it
Attack of the Eurogoogle (Need subscription).
No subscription needed here
From the article:
The disabled have been using voice-interface software for years. I might still have an old copy of Voicetype shovelware that came with a Pentium PC in 1995.
And for a search-engine-specific prior usage, in my area calling directory assistance has been largely automated using voice-recognition software, as have many coprorate phone systems.
Slashdot Burying Stories About Slashdot Media Owned
In looking at the description and the claims a bit, this is primarily an extension of their search interface. From the patent's description:
"...a method that provides search results includes receiving a voice search query from a user; deriving one or more recognition hypotheses from the voice search query, each recognition hypothesis being associated with a weight; constructing a weighted boolean query using the recognition hypotheses; providing the weighted boolean query to a search system; and providing results of the search system."
I can see how Google would want to use some form of voice recognition, defined as the 'shortest recognition hypothesis' or a combination of them, to convert it into text thereby using a search engine. Claim 20 could be read to mean a voice command converted to text for a computer readable 'search query' which is bound by the steps in the claim.
They're probably not going to own the idea of computer used voice generated searches, but they're seemingly moving to own voice command into Google searches as an added feature. Only as interesting as the voice recognition behind it...
CSorice
Working to make ideas into reality. www.i4e.com
between "a voice interface for a search engine" and a "generic computer voice interface program"? moreover, what is the difference between a "generic computer voice interface program" and a "generic computer interface program"? further, what is the difference between a "generic computer interface program" and a "generic computer program"? between a "generic computer program" and a programming language?
perhaps the only patentable thing in software might be breakthroughs in compiler theory or instruction sets, etc., because once you have a programming language, everything constructed from it could be produced by enough monkeys with typewriters, likely not even infinite. combinations (mash-ups) as patents? please, please no...
MORTAR COMBAT!
Now here's a thought - what if Goog were to implement their existing "Did you mean _____?" search suggestions with a voice-based system?
You speak "find Ly-nux!" - it speaks back "Did you mean Lin-nucks?"
While the example might make you chuckle, I think that such a feature could have far reaching impact - as has been discussed previously, one of the things which prevents people from learning about or discussing new things or ideas is an inability to pronounce the associated terms. Such an implementation would go a long way to solving that issue, in an emotionless and non-threatening manner.
There's a Starman, waiting in the sky / He'd like to come and meet us, but he hasn't got the time.
"If they don't patent it, somebody else will!"
And if I don't throw a punch at someone, they'll do it first... nice mentality
In Windows, it's fairly easy to hook up voice recognition to a browser. It would probably take me about 2 hours to come up with a basic voice-controlled browser.
As the article suggests, the vocabulary is the problem. When doing dictation, grammar rules allow the voice recognition engine to usually narrow the list of words making recognition more accurate.
The problem with searching is that you could be searching for anything. I'm not even sure their ideas of using past searches is even a particularly good idea. I do searches on such a wide variety of subjects, I'm not sure I'd want it to make assumptions about what I'm searching for.
This would be a nice application for a PDA or something, but for desktop, voice recognition just doesn't make much sense yet. The biggest problem is background noise, which really intereferes with recognition accuracy. Then there's the issue of whether or not you're actually talking to the computer. I like to listen to music when I work, so that's an issue, after all, it doesn't matter if it works for the rest of the world, it matters if it works for me!
There are various groups working on these problems and I suspect in the next 5 years, we'll start seeing much better voice recognition that can get past a lot of these problems. Once that happens, I think we'll start seeing a lot more voice recognition applications on the desktop.
All that said, voice recognition is in pretty good shape today. In a quiet environment, accuracy is pretty impressive. I've been playing with it off and on for about 6 years now and things have come a long way in that time. I remember when doing dictation had horrible accuracy and hogged my 200mhz Pentium. The extra computing power we have today definitely helps.
I think they meant Lamda Lamda Lamda. What a great movie.
Can I bum a sig?
These days it seems that legitimate companies need to apply for patents in order to cover thier asses... Either that or purchace a licence for the technology from useless patent holding companies like NTP to avoid getting sued.
I used Google voice search by phone... what something like 5 years ago? Really yes. It was one of their beta sites. You called in to the test phone number, spoke your query and then just hit search in your browser and the results came up.
I guess there were so few people testing it just pulled up the last voice print for the search. It was pretty cool though.
MjM
XKCD:Xeric Knowledge Comically Dispen
they have a demo (currently down) http://labs1.google.com/gvs.html
To try out this demo, please follow these simple steps:
1. Pick up the phone and call the automated voice search system at (650) 623-6706.
2. After the prompt Say your Search Keywords, say your query to the system.
3. Click this link and a new window will open with your voice search results.
4. Say another query, and the new window with the search results will be updated with the new results.
ERR 411[Max number of witty sigs reached]
Wasn't IBM or AT&T suppose to do something like this a long time ago? I remember a commercial showing some girl hum a few notes, and the search system would attempt to find the song/arist she was looking for based on those few notes.
This is my sig. There are many like it, but this one is mine.
So now I'll need to write a check to Google each time I yell "HONEY! WHERE ARE MY KEYS?" in the house?
I'm a fiscal conservative, it's a pity we don't have a political party anymore
Didn't Google Labs used to have a phone-in search project some years ago that didn't go anywhere? I'm pretty sure that they did, and looking at the date of filing for the patent, it's likely that's how it came about.
I wouldn't keep my hopes up for Google Voice Search any time soon based on that.
Christopher S. 'coldacid' Charabaruk -- coldacid.net
... Echelon. I hope we can combine it with this
Don't fight for your country, if your country does not fight for you.
Although there is a fixation on using speech to search normal web phrases, the larger implication here is using speech to search speech. Google has recently gotten into videos in a big way, along with voice chat. This tech would be useful to search inside the audio components there.
I would guess that the most immediately useful application of a voice interface to a search engine would be in facilitating searches from devices like cellphones. Right now, entering search terms in a cellphone is quite tedious. Another possible use of the underlying voice recognition technology would be voice to text conversion for chat in MMORPGs and other online games. Why? Because as voice enabled games and services like Xbox Live become more popular, the problem of "bad language" has continued to grow. Policing this is difficult because voice chat is not logged in any easy to search or filter manner, if at all. If Google lives up to its reputation, it's not a big stretch to see them building on their voice technology to provide voice chat to text logging for online game providers. Being able to log voice chat would also be useful to the gamers themselves.
To the making of books there is no end, so let's get started
My internet browser heard us saying the word Fry and it found a movie about Philip J. Fry for us. It also opened my calendar to Friday and ordered me some french fries.
Oh Shnap! This must be the secret Google-planned iPod killer! ...I'm sorry, let me put on my tinfoil hat.
Here is what I overheard at MS. "That was OUR idear! I'll FUCKING KILL THEM!" -Ballmer 2006
No sig for now.
People keep complaining to me how I want to do the job of the compiler, but getting the right phrase is the job of the speech recognition system, not the search engine, though I suppose the compiler is better at it's job than the speech recognition engine. Even so, I'd want to correct the speech recognition system for guessing wrong before I passed it on to the search engine, but such tools are poor, possibly for similar reasons people don't want to second guess the compiler.
No, I didn't say "Tonight to be sent him" I said "Sailing to Byzantium". Now, we're going to train on this phrase until you consistently get it right.
If I meant "free eye pod instead of "free iPod" I would have drawn out the eye and had a pause between eye and pod. Now if you used the canned results of a search engine to give the speech recognition engine a heads up as to how phrases are used, that's a whole different kettle of fish. Perhaps some variation "desktop search" should be used to tell the search engine how the user uses phrases, though currently "desktop search" doesn't distinguish between who wrote what.
Don't think of voice recognition on computers - that's so 1990's.
First off, I think patents like this are evil, and though I haven't read the details I suspect it's written broadly.
This patent applies to telephony and devices that haven't been invented yet. How slick would it be to integrate your browser on your cell phone with Google's services? Think about the integration with Asterisk you could do. You could have Google provide driving directions based on your spoken input. Want a picture of your favorite porn star? Say the name into Google and you could send a pic to your PDA.
The problem right now is speech recognition. Nuance is the best game in town and they're damn expensive. What Google (and the world!) needs is an open source speech recognition engine. (We're talking engines, not those cute, cuddly Naturally Speaking-like toys.) Then you can build server side apps relying on voice - say build it into every Asterisk server, into Vonage's network, etc. Those servers receive the voice requests and make the search query over the Internet to Google using a custom API. Google returns the result to them and they figure out what to do. This makes the infrastructure overhead negligible for Google and puts the burden of application development on others.
----- obSig
....covers a 'Voice interface for a search engine.' It says "A Search Engine", no names mentioned, so why are people assuming and coming to conclusions that it is google?? It could very well be other search engines
[Professor] Good news everyone... We have been Googled. Here's the numbers of individual hits. [Bender] Wohoo!! I win again. (singing) Bender is great, Bender is great ...
A start up company I once worked for already developed this idea, and proof of concept *working prototype*. The prior work provision would allow it to be an invalid patent if ever challenged.
Just like how Google is unveiling Writely, the Web Word Processor, which is speculated to be a cool and different from our routine MS-Word, so also, for the Mobile space it will be a totally different app. My experience suggests that its not just "e-", "v-", or "m-" nabling of technologies, it has to undergo some kind of transformation to be cool to use which ultimately makes it HOT in the market.
Finally, I don't think US is the best market for m-search & commerce. With just 207 Million subscribers in the US at the end of 2005, it is way behind the number subscribers in the East(Japan, China, Malaysia and India). The East and the Europe have worldclass 3G & 4G infrastructure with highly affordable subscription rates for the mobile savvy consumers. While for the internet era, it was the opposite. I guess Mobile 1.0/2.0 is a whole different game than Web 1.0 and Web 2.0.
Voice recognition doesn't work properly so why bother implementing something that is "broken" from the start? Why frustrate all those users who are going to have to repeatedly say their search term all because the software can't handle their accent? The keyboard may not be the best input device ever invented but at least typed words do not have an accent. Well, at least not when the words are spelled correctly.
That's all fine and good, but what I really want to do is go to meatspace.google.com and type in "Kia Sportage keys" to get white balloons floating over and pointing out keys that go to a Kia Sportage within an N meter spherical radius. Bonus points if the search algorithm doesn't involve a midget that lives under the porch, rifling through everything while you're not looking. When will this happen?
Help us build a better map!
http://agents.media.mit.edu/projects/voice/ is one way to increase recognition (this only scratches the surface compared to what Google could sematically do)
Google has jumped the shark.