Pages in topic:   [1 2] >
After machine translation, here comes machine interpreting!
Thread poster: Csaba Ban

Csaba Ban  Identity Verified
Hungary
Local time: 21:14
Member (2002)
English to Hungarian
+ ...
Nov 6, 2009

A new Japanese gadget captures voice, transfers it to a remote server for speech recognition, a software translates the text, formulates voice and sends it back to user's earphones.

http://www.telegraph.co.uk/technology/news/6493869/NEC-unveils-Tele-Scouter-translation-glasses.html


 

sivtufte  Identity Verified
Norway
Local time: 21:14
English to Norwegian (Bokmal)
+ ...
Useful if are articulated :) Nov 6, 2009

It will be of use for household amusement but no doubt a great developement - if you don't have an interpenter...

However, the words that read the same may get in the way - hopefylly not as badly (or amusing) as in this service;

http://www.translationparty.com

I use this if I find work hard or impossible - to remind me of how great I really am!
Enjoy it.
icon_smile.gif


 

xxxL.G.F.  Identity Verified
Spain
English to Spanish
+ ...
Machines will never make our hard and valuable job Nov 6, 2009

Well, I am really shocked after reading that article from the Daily Telegraph. I do actually think Japanese are obsessed with the idea of inventing a device substituting our job as translators and interpreters. Please, pay attention to the following extract from the article:
"The Japanese manufacturer admits that the device's translation capabilities are limited at the moment (...)"
This type of things makes me laugh, it is impossible a device has the same capabilities than human beings. We cannot forget wonderful automatic translation machines, absolutely unuseful (I am not going to give you any examples, I would not like to be sued for thaticon_smile.gif .)


 

Marek Daroszewski (MrMarDar)  Identity Verified
Local time: 21:14
English to Polish
+ ...
CAT vs. Human translation Nov 6, 2009

The CAT vs. Human translation topic has been discussed at length, but not put to the test AFAIK.

Now you can try for youselficon_smile.gif



Using the above link try translating:

All CATs can do much better than humans.

Have fun!
icon_smile.gif
Marek


 

Williamson  Identity Verified
United Kingdom
Local time: 20:14
Flemish to English
+ ...
Robotics Nov 6, 2009

Lucas González Fernández wrote:

Well, I am really shocked after reading that article from the Daily Telegraph. I do actually think Japanese are obsessed with the idea of inventing a device substituting our job as translators and interpreters. Please, pay attention to the following extract from the article:
"The Japanese manufacturer admits that the device's translation capabilities are limited at the moment (...)"
This type of things makes me laugh, it is impossible a device has the same capabilities than human beings. We cannot forget wonderful automatic translation machines, absolutely unuseful (I am not going to give you any examples, I would not like to be sued for thaticon_smile.gif .)


Never say never. The Japanese are more advanced in robotics. A robot is device that has some human capabilities. Those who laugh now, will laugh less within now and say 15 years, but then I will not be my problem any longer. Imagine how much international institutions can cut on their budgets if such tools become more efficient. Where was the pc in 1980: 16 Kb member, no Microsoft,...
Where is the pc now?


 

Claudio Porcellana  Identity Verified
Italy
machine interpreting ..... Nov 6, 2009

I guessed as much! ....
now even interpreters have their share ...
LOL

but why stopping only here?

software companies can even create a MP (machine proofreader)

so, a machine will translate the text and another machine will proofread it ... GREAT!
finally a robot will read its (his ?) own robot's manual

abdicate completely to machines and you'll be done for, mankind !
;-(((

Claudio


 

Pablo Grosschmid  Identity Verified
Spain
Local time: 21:14
English to Spanish
+ ...
it IS possible, and maybe sooner as we would like Nov 6, 2009

Williamson is right, it is just a question of time.

After four decades in interpretation and translation, I dare say that we will probably see a "usable" interpretation machine much before a usable text translation machine sees the light.

The law of exponential growth in computer power means that a machine already is, or will be very soon, capable of accomplishing any mental task currently carried out by humans. It just has to be fed with the right stuff. Satisfactory text-to-speech and speech-to-text programmes already exist and are rapidly improving. Only the internal core engine between them is lacking, and the infantile "grammar-and-dictionary" based text-MT systems to make gadgets (like this new japanese thing) are not good enough even for a start.

But, alas, almost nobody, including most interpreters, really know how human interpretation functions. Thus, until now nobody has thought of modelling into a computer engine and replicate the mental processes applied and the information used by human interpreters (which, by the way, are only half-explored). This a huge task, but POSSIBLE.

Once a big company realizes which is the way to go, the development can be very fast.

Sorry for the bad news.
















[Edited at 2009-11-06 21:39 GMT]

[Edited at 2009-11-06 21:39 GMT]


 

Claudio Porcellana  Identity Verified
Italy
blind trust in technology ... Nov 6, 2009

... seems unbelievable to me!
mankind is wasting from centuries a lot of energy/money to redo (badly) what nature does very well effortlessly

for example, I recently read the story of sericulture (natural, technologic, hyper-technologic and then natural again) and I found it very illuminating ...
;-D

I think that science/technology should try to fulfil its old and vain promise, i.e to free men from physical struggle, rather than trying to substitute our cleverness, that is insane

Claudio


 

Vlatko
Local time: 21:14
English to Macedonian
+ ...
Short and simple? Nov 7, 2009

Short and simple. Not machine, not software, nothing on this world will never work untile human eyes and hands don't manualy push it and push it to do the work as it should be done.
Machines that cost 600 000 eur. are not able to reach the purpose.
Every developing leeds to easy way for job to be done, but it will never exclude the human sense.
Where did PC development get? From 1980 to today, nothing is new, it just plastic/sand machines have become smaler in size and cheaper, available to any home = competition gets high = better results for less money.
No one will never be able to create such translation base for automated translation with quality, it will take 1 000 000 translators from whole the world and 10 years to fill it in any sentence combinations. = Even Google + Microsoft together are not able to invest such money. = If some one do it, the service will cost more expensive than human brain+hand.
Google, yes, continue to translate web pages for simple people, it's no need for quality for them???


 

Williamson  Identity Verified
United Kingdom
Local time: 20:14
Flemish to English
+ ...
NEC Nov 7, 2009

Technology evolves at a rapid pace.
Nippon Electronics Company happens to be one of those companies engaged in robotics. If I am not mistaken, it created female robot, looking like a human. The robot has a speech module and reacts to sentences from humans. Next step, program that robot so that it (s)he repeats/answers in another language. In other words, interpret.
Not for now perhaps. NEC is a multinational, Google and Microsoft are not small comapnies either. Look at it from an R.O.I. point of view. How much would it cost and how much would it save if interpreting/translation by humans could be minimized (or even eliminated).
Machines costing 600.000 euro. How much is the budget of the E.U. is a question which occurred at a preselection test of an EU-competition: 126 billion euros was the correct answer.
Besides in 1982, I wanted to buy an Apple Lisa II comptuer with 32 Kb and AppleWriter (elementary word processor) for the equivalent sum of €2000. What do you have for that kind of money now? In 1994, I bought the first cell phone (Moterola) for €450, the price of an Iphone today. Ever heard of the BCG-Matrix: star, cash-cow, ?, dog or the life cycle of a product: early adopters pay a lot, mass consumption takes place and prices go down.




[Edited at 2009-11-07 08:34 GMT]


 

Claudio Porcellana  Identity Verified
Italy
technology Nov 7, 2009

BTW, the technology that took men on the moon was very simple (something as 2 or 3 Commodore 64 and many piece of foil over the LEM)

After 30 years, PCs became a lot more sophisticated and in spite of all, there are no human colonies on the moon and EU HCL was blocked (again) by some bread crumbs ...

Technology doesn't ever advances
Often it fails and it will fail in this case because the the game's not worth the candle

It is an existential issue, not different from the agricultural revolution

Even there, scientists guaranteed that introducing machines, the peasant's life should have been made easier

Now, we see that it was a lie: current peasants work a lot more and harder than their grandfathers (even if the manual work is apparently reduced), as they have to pay much more to maintain more and more bigger, expensive and sophisticated machines, plus to buy fuels, pesticides, etc

In the meanwhile, machines created a massive unemployment and the shift of young people from countryside towards the cities, i.e. the mother of ambiental degradation and countryside/mountain abandonment

Is was really an advance?
I don't think so

Claudio



[Modificato alle 2009-11-07 09:08 GMT]


 

FarkasAndras
Local time: 21:14
English to Hungarian
+ ...
??? Nov 7, 2009

Pablo Grosschmid wrote:

After four decades in interpretation and translation, I dare say that we will probably see a "usable" interpretation machine much before a usable text translation machine sees the light.



Why do you think so?
I find this a laughable statement, to be honest.
For interpretation, the sw has to do voice recognition first, and then feed the text to the same algorithm that the written MT uses... the extra errors introduced by the voice recognition will always be there and make machine interpreting worse than MT. Add to that the issues with voice synthetizing and the time pressure and you see why interpreting will always lag behind.

Depending on your definition, usable text translators do exist. The top current MT software is easily good enough for you to make some sense of a newspaper article written in a foreign tongue, and some of our colleagues are using such sw in their day-to-day work.


 

Parrot  Identity Verified
Spain
Local time: 21:14
Member
Spanish to English
+ ...
Not to be taken lightly Nov 7, 2009

Pablo Grosschmid wrote:

But, alas, almost nobody, including most interpreters, really know how human interpretation functions. Thus, until now nobody has thought of modelling into a computer engine and replicate the mental processes applied and the information used by human interpreters (which, by the way, are only half-explored). This a huge task, but POSSIBLE.


I've read some of the models proposed. EXTREMELY complex. To begin with, interpretation doesn't function literally.


 

FarkasAndras
Local time: 21:14
English to Hungarian
+ ...
Modelling human processes Nov 7, 2009

Pablo Grosschmid wrote:

But, alas, almost nobody, including most interpreters, really know how human interpretation functions. Thus, until now nobody has thought of modelling into a computer engine and replicate the mental processes applied and the information used by human interpreters (which, by the way, are only half-explored).


Probably true, definitely beside the point.
Computers work differently from human brains so programmers make them approach problems differently from what a human would do. It would make no sense to make them imitate humans because they are bad at it, and some of their amazing capabilities (recalling data instantly with perfect precision from huge databases) would remain unused.
For example, computer chess programs don't play chess quite like humans: in endgames with 6 or less pieces on the board, they don't make plans, they don't calculate how vulnerable the opponent's king is or what the most useful position for the dark-square bishop is. They have a massive pre-made database that tells them what the best move is in the given position and they play it instantly. Humans can't possibly remember that much data of course, so they are stuck analysing.

MT itself is a great example of how massive the difference can be: humans make sense of the source language text and then formulate it in the target language. Computers can't really do that and I'm not sure they will ever be able to. They attempt to come somewhat close with one of two the main MT algorithms, rule-based MT: they identify grammatical structures and recreate them in the target language, while slotting in the translations of words. They don't "understand" the text even with this process, but the MT method that's currently most successful is entirely different: statistical machine translation basically uses fuzzy matches from a gigantic TM and adapts the match to the source segment.
This has hardly anything to do with how humans translate/interpret but it's the best way for machines to do it.


 

Williamson  Identity Verified
United Kingdom
Local time: 20:14
Flemish to English
+ ...
Worries for the young ones. Nov 7, 2009

Some who are 25-30 years old should better start worrying.
If a career, freelance or otherwise as an interpreter spans 4 decades, where will they be in say 2050 at the current rate of technological evolution. I know where I will be in 2050 : dead and gone. No worries for me.
Make no mistake, if the current Mekka of interpreting (E.U.-institutions)can save costs by gradually replacing interpreters by machines (robots) which are (almost) as perfomant as humans, it will be done.
Not by sacking people, but by not replacing them by youngsters. That's the way social minded state owned companies do it.


 
Pages in topic:   [1 2] >


To report site rules violations or get help, contact a site moderator:


You can also contact site staff by submitting a support request »

After machine translation, here comes machine interpreting!

Advanced search







SDL Trados Studio 2017 only €435 / $519
Get the cheapest prices for SDL Trados Studio 2017 on ProZ.com

Join this translator’s group buy brought to you by ProZ.com and buy SDL Trados Studio 2017 Freelance for only €435 / $519 / £345 / ¥63000 You will also receive FREE access to Studio 2019 when released.

More info »
Déjà Vu X3
Try it, Love it

Find out why Déjà Vu is today the most flexible, customizable and user-friendly tool on the market. See the brand new features in action: *Completely redesigned user interface *Live Preview *Inline spell checking *Inline

More info »



Forums
  • All of ProZ.com
  • Term search
  • Jobs
  • Forums
  • Multiple search