Real time voice translation is on the way and Google aims to pioneer it.
The Internet giant wants to develop its smartphone technology to translate speech in real time.
The company would combine its advanced voice recognition know-how with its text translation service to create a mobile phone that acts as an instant interpreter.
Head of translation services Franz Och said: “We think speech-to-speech translation should be possible and work reasonably well in a few years’ time.“Clearly, for it to work smoothly, you need a combination of high-accuracy machine translation and high-accuracy voice recognition, and that’s what we’re working on.”
‘Everyone has a different voice, accent and pitch,’ said Mr Och. ‘But recognition should be effective with mobile phones because by nature they are personal to you.’
Google has already created an automatic system for translating text on computers, which is being honed by scanning millions of multi-lingual websites and documents.
However Google admits speech will be an even tougher challenge than text but says a customer’s phone would adapt to its user by ‘learning’ their style of talking.
Source: The Times
Well, Google can't be pioneers as Interlecta have been doing this for years on multiple mobile platforms.
it doesnt translate speech so what are you talkin about
Russians can not into technologies
I wish Google well but, as a linguist, I'm not convinced that they are aware of the enormity of the task that faces them. I advocate a non-technological solution, i.e. wider use of the planned international language Esperanto. At least we know that works.
As a professional literary translator, I can tell they have completely misunderstood the nature of human language. It is not very logical (perfectly logical languages may be good enough for computers, but not for people – goodbye, Loglan!), and it’s basic unit is not the word but the phrase; and constructing a complete list of phrases just isn’t possible. Esperanto seems to have found the best balance in so far between logics and humanity.
Wow.
I look forward to hilariously mangled voice conversations.
Wow, is Google amazing or what. I think Google should run for President.
Jess
http://www.private-surfing.be.tc
Maybe, with them developing this type of technology, they should look at acquiring babelfish.
im excited about this!
http://doublequoted.com
It sounds like a tough task…but I have two words for you…Street View. If they can do that….
“Now it is such a bizarrely improbable coincident that anything so mindbogglingly useful could have evolved purely by chance that some thinkers have chosen to see it as a final and clinching proof of the nonexistence of God.”
“The argument goes something like this: ‘I refuse to prove that I exist,’ says God, ‘for proof denies faith, and without faith I am nothing.’
” ‘But,’ says Man, ‘the Babel fish is a dead giveaway, isn’t it? It could not have evolved by chance. It proves you exist, and so therefore, by your own arguments, you don’t. QED.’
” ‘Oh dear,’ says God, ‘I hadn’t thought of that,’ and promptly vanishes in a puff of logic.
I wish Google well and, as I Google Voice user, I hope they achieve their goals. But considering how mangled and inaccurate the transcribed Google Voice messages are, I think we have a long wait until this is ready.
fuck
All I have to say is good luck with that! Have you ever called a company which uses voice activation/recognition whatever, they cannot even understand the English language much less work with other languages and dialects.
Sounds too good to be true but its long ahead us. The email translation is still primitive so imagine…
Where are some working examples of usage of Esperanto? I'm not aware of any.
http://jimasks.me/could-you-fall-in-love-with-s…
Are you cunning?
That's cool! I had that idea about 8 months ago. I even specified google translate. http://bit.ly/5b3N72
Just use english and STFU!
Try the 'Internacia Kalendaro' of meetings using Esperanto almost on a daily basis around the world right here, if you don't believe it!:
http://www.eventoj.hu/
or listen to the daily Esperanto blog from Radio Polonia:
http://www.polskieradio.pl/eo/
Well, here's one example (chosen at random — because I was looking at this particular Esperanto-language Wikipedia article about the Alaskan Goldrush quite recently):
http://eo.wikipedia.org/wiki/Alaska_orimpetego
Here's another, from the Esperanto version of Le Monde Diplomatique,
(again, just because it's in my browser history already):
http://eo.mondediplo.com/article1566.html
The regular Radio Poland Esperanto podcast:
http://www.polskieradio.pl/eo/
And a Polish Esperanto Youth-group's web-site
http://viavento.republika.pl/esperante.htm
and their regular podcasts and videos:
http://www.podkasto.net/
http://viavento.republika.pl/aktuale/vvfilm.htm
The Esperanto translation of Darwin's “Origin of the Species”
— the online version (much easier to read than Google Books):
http://www012.upp.so-net.ne.jp/klivo/origino/en…
That should be plenty to give you an idea. And we haven't even touched on fiction, literature or poetry…
Google's “Babel Fish” translator will in never solve the language problem. Not only does it discriminate against anyone who cannot afford a mobile phone, but against minority language groups as well.
There are 6,800 languages worldwide, not fifty-two !
Moreover, if I met a native in Borneo, and he said to me in Hakka “I've lost my mobile phone” how would I understand him :)
And how many starving Africans can afford a mobile phone !
As English loses its economic power, the answer is not for us to move to Mandarin Chinese, but to Esperanto which puts all speakers on an equal footing.
Have a look at http://www.lernu.net or http://www.esperanto.net
> about “enormity of the task”
To tackle such a difficult problem, one usually builds a model that is able to solve a simpler problem, and one gathers experience from that model.
They would learn a lot by trying to translate from Esperanto to English.
1) Esperanto is a language that is easier to analyse, so the main difficulty will lie in the second part of the translation process.
2) If point 1) works satisfactorily, one can then try with another target language.
3) The next step should be to translate from Esperanto to English.
4) If it appears too difficult to find a good solution for another source language, we will still have something that works with Esperanto.
5) Even if the translation to Esperanto does not give satisfaction, it won't be too difficult for a human to correct it, and produce a text that can be automatically translated to other languages (assuming step 2) was realized).
If translating from and/or to Esperanto proves too difficult, then it is hopeless to attempt to translate from/to another language.
There is a lot of documentation about that under google:”DLT Distributed Language Translation”
What seems to have a chance of success is the “brutal force” approach. It can succeed in situation when the corpus is limited to a particular domain and there are a lot of translations available already.
Remuŝ
Some further thoughts:
Is this really the answer? Or even a good answer?
This sounds to me like another case of trying to implement a technological solution to a non-technical problem. It's worth asking whether this is really the right approach.
How well is this going to work trying to chat up a girl in a foreign bar, on a foreign beach or on the bus?
How well is this going to work to negotiate an informal agreement, striking a deal in the corridor, or mutually feeling out one's opposite number around the coffee-service?
And how sure can you really be that the technology has accurately conveyed the meaning — let alone connotation and nuance — you wanted to get across?
It should also be considered how well this is — or isn't — going to allow for privacy, in a world where e-mail is already routinely scanned and indexed by both commercial and government entities (including Google itself, and the US and the Chinese governments, just for starters)? Would you trust this system when negotiating an important business deal, consulting on legal matters, etc?
Perhaps something like Esperanto — which has been around for well over a century and shown to work quite well in practice — deserves a fresh look.
Translation software has come a long way in the last few years so I'm not surprised to hear about this. Between some languages it might actually work reasonably soon.
Not actually as difficult as all the nay-sayers above believe. There’s already a huge library of people who have read books (e.g. librivox). Take the text, take the recordings and use them to test and refine the existing voice recognition software.
And who actually cares if the translation is a little off the first time. I doubt many of you actually speak second language, or a third, or fourth. I speak 5 to differing degrees of fluency and even two fluent speakers have to rephrase stuff to make a point clear and understood. So what if the first translation gets a puzzled expression, just repeat it a different way.
As for the useless git who whined about there being thousands of languages out there, get with the program. It’s either develop a useful real time voice translator OR get everyone to learn a new common language (and all those Esperanto fans it’s NOT going to be Esperanto since more people speak English and it’s already the language of business, so drop the entire Esperanto thing, it’s a dead duck). If you want your language included then get recording and stop whining.
Google has a little thing out called Google Voice. On Google Voice when someone leaves you a voicemail the system automatically converts the spoken word to text. As the recorded voice of the person who left the message speaks each word is highlighted and underlined at the same time it is spoken, like when a kids song on the television and as the word is being sung the printed word is highlighted on screen. It is very amazing how the technology works and it fully blows my mind that it works so well. Next to each message there is the option to click either a Check Mark if the speech-to-text translation was good and an ‘X’ if the speech-to-text translation was bad, after you choose whether it was good or bad a box pops up and asks if you would like to donate these results to Google. What I am imagining is that since everyone speaks differently and the system has no idea what kind of accent or speech pattern or impediment the caller will have is that they will use these donated results to fine tune the task of mapping human speech from all walks and types of life and use it for example as at least a starting point for something like this translation service. I know it seems like a huge task to most people, but Google has unimaginable resources and the can accomplish any task they deem important enough to embark on. Real time translation will be as important as cell phones themselves within 10 years from now. Mark 2020 on your calendar. It will save lives when everyone can speak easily with anyone!
Google is impressive in its capabilities however our little company already has real time voice translation that is accessible on any phone. No need for a smart phone, or even a cell phone: http://lexifone.com/