Whoever created this deserves the boatload of money they are going to make. It has the potential of radically change the experience abroad for people traveling in foreign countries.
If this thing can be made to do Chinese or Japanese... even if it does so haltingly, with mistakes...
Raise the price. For the love of god. Raise the price. This is much, much more valuable than five bucks. There have been times in life where I'd gladly have paid a dollar per minute for this. And I'm a cheapskate.
Raise the price. For the love of god. Raise the price.
Absolutely, read http://bit.ly/bdHKmm to understand why. (Sorry for the shortened URL, the URL recognition software here doesn't understand URLs with apostrophes in here, so it actually is necessary.)
Yes. No Android yet but this seems worth $10 to me. I'd do Chinese first. There's a big market for that in California; many Asian families have a language gap between grandparents or parents born abroad and a younger generation born here. Basic communication is easy but things slow down fast if conversation turns to anything complex. Being able to print some words neatly by hand in English and instantly get Cantonese would sell like crazy - I have 5 in-laws for whom it would be a no-brainer.
We have a lot more Vietnam refugees here than we do Chinese folks. If you'll notice, the three main languages you can get government papers in (at least over in San Jose, no experience elsewhere) is English, Spanish, and Vietnamese.
San Jose is a pretty skewed example as it has a very high Vietnamese population, where over California there is probably an order of magnitude more Chinese immigrants.
Vietnamese would be easier to recognize/do a translation pack for too, now you mention it - basically the Latin alphabet with extra marks to make 30 letters. Apparently there are more north Vietnamese around here and Oakland, more South Vietnamese in SJ and down south, maybe 450k in all of CA. Chinese-American maybe 5-600k? More than 150k in SF.
Incidentally, there's other and easier untapped startup opportunities for the Asian language demographic, if anyone wants to send me a note via gmail
My in-laws are a mix so either language is good, both is better. I can promise 3 Android and 1 iPhone sale just in my home :)
If I was traveling to japan or china, I would gladly pay $50 per language to translate. That's probably too high for an in-app purchase but I think $29 would be a no-brainer for languages that don't use English characters.
The dictionary is free and the OCR is available as an in app purchase for $15.
I'm just saying this because it is not trivial to adapt an OCR engine to non-latin scripts, because the image analysis techniques are rather different.
As I was suffering from acute future shock last night I didn't make this clear... but I'd happily pay a premium for French, a much easier OCR problem (and, indeed, one of the easiest foreign languages for an English speaker to read, since there are so many loan words in both directions.)
I understand that Chinese is harder, for many reasons. It is also much harder for humans. But that is why I'd happily pay, say, $10 per day of my trip for an engine that gives reasonable hints. Maybe more. Try me and see.
Of course, once this technology is ubiquitous and the price of such a thing has fallen to nigh-zero it is going to change the world. But we have to walk before we run, and you'll need the money for R&D and legal costs.
Absolutely. I travel pretty much nonstop and I would pay dearly to be able to translate signs on arrival into a new country. Things like 'ATM' and 'Toilet' are crucial.
That's very, very impressive. Traveling would be so much easier with just this app, amazing! Let's hope they'll support Asian languages in the future, for round-the-world coverage.
The awesome thing is that once the platform is there, they could easily open it up to other languages and get them rolling too--though they might have to really think before diving into languages with a different character set, such as Japanese.
Very impressive. I'm curious to see how it performs in the wild though -- can it translate signs outdoors just as well as in a studio? I really hope so.
It performs very nicely in the wild. One of the nice things about the "real time" aspect of this app is that you can nudge the phone around until you get a good translation.
What about smaller type (and sometimes weird fonts) in somewhat dim environments – I’m thinking of menus, that seems like a typical situation where translation is necessary for tourists.
I've known about this application for a long time, Otavio showed me an early prototype of this running on his laptop over a year ago. I'm ecstatic that they finally released it.
(I'm also happy that I was able to get their story on HN before TechCrunch :D)
If Smartphone + Wikipedia + Google = Hitchhiker's Guide to the Galaxy, then this is the Babel fish.
And I’m sure that it will also, by effectively removing all barriers to communication between different cultures and races, cause more and bloodier wars than anything else in the history of creation.
(hola, i'm john deweese, one of the creators)
good to hear the early positive response, and we look forward to your insightful feedback (support@questvisual.com) -- mention this site so we pay some extra attention.
i hope this app will be nice and disruptive, and we'll be looking carefully at what people expect and how they actually use it. it's a platform, and we're really excited about the directions it opens up.
An android or SDK version that could be used with a heads up display system would be impressive. If you can do more then Spanish, you should consider talking to the US Military.
For the love of FSM, do it! I speak for a vast majority of Android users here on HN when I say I'd buy this even if it cost five times what it does now.
From the technological point of view I can just say wow! Congratulations, this is a great app.
Aside from that, and as a Spanish native speaker, the translated text sounds really weird to me. Even without knowing English I would be able to understand the translated text (at least the example that you gave) but it can be improved. But this is the only con that I see to your application. Probably you're already working on that, but I thought that you might like this input.
spot on! I can usually muddle through Spanish and French signs when I travel, but I'm terrified of traveling to China or Japan where the signs are quite a bit more impenetrable.
Depending on how their OCR works, Mandarin/Japanese->English(/spanish/&c) could be a _lot_ more difficult (thousands of characters to distinguish) and in the case of Japanese the translation a lot trickier (completely different grammar, Mandarin's not so bad though).
This could just be a case of needing to optimize and train their recognizer to deal with a much larger set of possible characters, or it could require implementing kanji-specific OCR techniques like attempting to decompose the characters into their constituent strokes, and recognize based on classification of those strokes (orientation, position, direction).
This is what I was thinking. I don't know about Japanese, but sentence structure in mandarin is pretty much reversed from what it is in romance languages. And so much of character meaning is based on context that outside of common phrases it might be nearly impossible.
On the other hand, an English to Chinese translator might be more doable, and might also be more commercially profitable.
Even an app that converted characters into the most direct meaning could provide insight in a large number of cases into the intended meaning and would be valuable.
An app that converted from characters to pinyin (a much easier problem) would be gold to someone trying to learn mandarin. I'd easily pay $50 for something with that functionality without it even making an attempt at the english. It's a much smaller market than those trying to understand chinese signs on a vacation, but it's one with a much larger stake and interest in the result.
I would have absolutely no problem reading Japanese or Chinese if it just did a word-for-word translation, romanized all the particles, and expected me to know what they meant (or had a handy feature that explained them when the image was paused). Even so far from actual translation, this would still pass for one of the best things ever.
Of course, guessing the correct word when performing word-for-word translation of hanzi is almost impossible, so even the extremely primitive product I'm thinking of is very difficult.
You should register wordlensapp.com or something of the like. I'm not sure hosting nothing but the app promotion on a site with a completely different URL is beneficial.
With such a product, you should perhaps brand yourself more on your product than your company name. :)
There are many trends for using domains similar to what you would desire: get<product> being the best, and <product>app being one of the other choices.
I have very bad experiences with finding an app's site, if it models <product>app. Maybe it's just bad SEO on their part, but I imagine that it'd be an awful hassle for any normal user who don't know the conventions.
Then it struck me that <use>product would be a very interesting domain that makes much more sense than <product>app. It also mirrors the imperative <get>, although it may not be as popular and known to users - yet.
I think it's a shame they didn't secure a regular <product>.com domain for such a great product - but, on the other hand, I'm sure all the press will forgive them (hell, they're trending on Twitter, and 80% of my social media digest today was about the bloody thing).
I've registered usewordlens.com, usewordlens.net, and usewordlens.org and will be happy to hand it over or point it to a domain of their choice (if someone tells me how the hell to do it using name.com, because I'm a complete idiot in that regard. I guess I have to mess with some DNS).
Since you mentioned this, I registered wordlens.co and set it to redirect to their website. I emailed the info and asked them if they want it. That's how awesome I think this app is - I'd rather spend my own money to make sure they get a good domain.
otaviogood or johndeweese, if you're reading this, my email is in my profile.
You guys have made Doctor Who's Tardis translation-in-the-brain come one step closer to being real. I just tried it and it works perfectly. In my minimal experience, it works great on medium sized text so you may want to advice users to get closer/further from the text if your app can't find anything to translate (i.e. OCR fails).
As a language learner I'd love the ability to capture and export (in some fashion) the pre-translated text, as text not an image, to import into SRS learning software. This especially applies to the Japanese/Mandarin support that I hope you guys are working on.
1. The logo seems similar to the United Nations logo. What was your inspiration for the logo?
2. In the video, who is holding the camera? And who is holding the signs?
3. Did you use an ad agency to make the video, or did you put it together yourselves?
The first card says "welcome to the future" which I thought was perfect :)
The designer who made the logo said he was going for a passport look. My brother made the video. He also did the music in the video. He's pretty good with that stuff.
The only thing I noticed is that it translates the text to all caps, and the slides were all caps, so you didn't notice anything in the video. But that's really not material; to obsess over that is missing the point that it translates text in real time between human languages using a camera.
It’s also free (not the translation part but the word recognition and replacement part) so there is no excuse to not try it out yourself or getting a friend to try it out for you (should you lack an iPhone).
The color matching is easy. It's not using the same fonts, and pretty obviously stretching the text to take up the approximate amount of line space - also easy. Don't be too quick to judge! :)
I downloaded it immediately. Making it free to try and selling the dictionaries is a great sales model. The demo mode does a mirror-image reversal of the word to show you that it can detect words.
I was using a 4th gen iPod Touch, so lower-res camera and no flash.
Unfortunately, I was not able to get it to work as well as I'd hoped. I tested physical copies of a few things:
1. Mac OS X Snow Leopard cd case. Black sans-serif text on white background. This worked the best, with the word-reversal consistently getting "Snow" and "Leopard" reversed with little flicker. "Mac" seemed to go in and out. OS was rarely replaced, and usually along with "Mac" as though it were part of the same word.
2. C++ Programming language book cover (http://pixhost.info/pictures/631454). Dark blue serif text on white. This almost never worked. When it did recognize letters, the recognition shifted so much that the word was a constantly moving jumble.
3. Throat Coat tea bag. White serif text on Red. At any point in time, about 50% of the words were recognized and reversed.
You can take a snapshot, and each word it recognizes is highlighted, which is pretty cool.
I would definitely buy something like this that handled non-romance languages to English.
The current iPod touch has only a very low res, low quality camera (I think VGA or maybe a bit better, in any case less than 1 MP) without even autofocus. It would be nice if someone could try it with an iPhone to see if it fares better.
For me it seemed to work better in the translation mode. The reversing mode flickered forth and back, but with English-to-Spanish seemed to be more stable.
And I see the opposite. Anyone could have had this idea, but very few could execute on it. In fact all that will determine whether this is a big success or not is whether the thing really works or not (execution).
No kidding... I was just sitting down to research open-source Java OCR software to do this exact thing on Android. My dreams are dashed but I'm glad someone did it...
A lot of people must have thought of making OCR iPhone apps. But the idea of using OCR to do real-time language translation superposed on the original background image? That's genius. I hadn't thought of it. Did you?
At least in my limited experience, deciding what to work on is really important.
Acknowledging great ideas doesn't take away anything from their superb execution.
I did: "One that I've mentioned in almost every post here for ages, a service that I can take a photo, it does OCR and feeds the text into Google translate. Bonus points if it does source-language-detection. More bonus points if it is actually usable in a foreign country on real live things like signposts, menus, timetables, advertisements, book covers, leaflets, etc." - me, http://news.ycombinator.com/item?id=569838
Nearly two years ago, travelling in a foreign country with an iPhone, knowing Google Translate existed, knowing OCR existed, and not being able to do anything with that knowledge.
(It's so rare that I hit this kind of advance view on things I'm going to be shamelessly happy about it. NB, I'm not claiming anything here - I couldn't write it and didn't try).
My work colleague also had this exact idea, and told me about it in the hopes I could code something to do the job. There was just no way I had the time/resources/skill to make something like this.
I disagree. There is a comment higher up suggesting that this app could change the way people live their lives. I read this and it occurred to me that this was likely true. With the caveats if they market it correctly, if they handle competition correctly, if they expand and the proper pace, etc.
The creators will likely see a payoff from this whether or not they do any of these things correctly. But the magnitude of that payoff is entirely dependent on those ifs and the difference between executing correctly and not is humongous.
A friend of my girlfriend is part of the team that made this and he gave me a demo of this 5 or 6 months ago and even back then when it was in an early alpha form I was absolutely blown away by this thing.
Seeing this thing in action in real life is life changing.
I agree. I saw a demo at Epicenter cafe randomly by this guy about 6 months ago. Its one of the only demos I've ever seen where I felt like 'holy shit I've just seen the future'.
The military doesn't have to make a bid to buy them. All they have to do is throw down a billion dollar contract for prototyping an arabic version and Quest Visual will have no reason to sell.
[EDIT: Note, Translating Muerto Fin to English in Google Translate, does result in 'Dead End'. Can any Spanish reader clarify why the original Google Translate chose to translate 'DEAD END' to 'callejón sin sali'.]
[EDIT: So let me get this straight. Their program, running on iPhone [256MB RAM, 600 MHz ARM CPU], can take a live image, perform OCR, translate, create another image with the translated words. And all of this happens real time? Wow.]
I think it's really hard to get a 'Wow' reaction from HN crowd. And you have hit a home run! As someone else said in this thread, you are going to make boatload of money. Congrats!
Thank you. I think context is extremely important and very difficult to get right. I remember reading one article by Google engineer where they discuss how they keep improving their analysis of search queries.
E.g.
[Search Term] -> [Interpretation]
New York -> New York city
New York Times -> The news paper
New York Times Square -> Famous tourist spot in NYC.
There were other examples in the article. I will try to find that article and link it up.
I think Google should buy these guys and provide them with their knowledge of 'context'. Google Translate team and these guys should talk right away.
I used to work in machine translation, about 15 years ago. At the time, one of the big things they were working on was an improvement to the contextual inferences. The prior versions were able to consider context within a single sentence, and input documents could be tagged with a subject area, so the system would know to resolve ambiguities in favor of, say, medical terms.
The best examples of this, I thought, were the following passages:
Yesterday I went to the symphony. The sound was beautiful.
versus
Yesterday I went to Long Island. The sound was beautiful.
In the first passage, sound refers to an auditory experience; in the second, it refers to a body of water. Within the sentence "The sound was beautiful", there's no way to know which to use, so one must look to surrounding information to make a guess.
But I've given a short and clear example; anyone can see the right solution. But in real-life usage it's frequently unclear.
But I think that this app is doing almost a word by word translations, without a list of usual expressions or grammatical analysis.
For example, the correct translation of 'Dead End' is 'callejón sin salida', that means literally something like 'street without exit'. Translating word by word 'Dead End' you obtain 'Muerto Fin' that is unintelligible.
Another example from the video is the translation of
'Lengua boliviana con una salsa picante de anchoas'.
The app translates this as
'Tongue Bolivian with a sauce spicy of anchovies',
But the correct translations is something like
'Bolivian Tongue with a spicy sauce of anchovies',
because in a translation between English and Spanish you must reverse the order of the adjectives and nouns.
yep, this is a word-for-word translation, because it's fast and it gets the point across. we're working to improve translation quality and finesse, but it's a much harder problem to understand grammar. so, we hope it gives the general meaning, and you can learn to piece it apart.
Word for word is far, far better than nothing. And no network connection required, so awesome. Roaming charges when travelling are brutal, so very important feature.
Real time OCR + word for word translation....maybe its not particularly technically interesting on paper, but to watch it on a handheld device in front of your eyes is breathtaking.
Do you have a timeline for other languages? (Japanese)
Yeah, no network connection is the killer app here, since you're probably going to need this in another country where you probably don't have network access.
I think that most of the time a word-for-word translation is enough. I use Google translator a lot, and most of the time the translation is not perfect but it is good enough. So I think that is app is very useful.
But I think that the real problem are the idioms and phrases, like "dead end", that have a completely different translation.
On the other hand, I think that you have more processing time that 1/10 sec. Most of the time the user will point the app to the same text for 10 or 15 seconds. I think that it is possible to show first a very fast translation almost instantly and a few seconds later show an improved version.
I know that Google Translate is not word-to-word, but it is not perfect.
For example, a few years ago, some students of my wife give to her a homework about clocks and gears. When she read it, she was annoyed because the redaction was incredible horrible. But later we realized that the students didn't write it, the "homework" was a web page translated with Google Translate.
Another time, I need an example of the differences between JPEG and JPEG2K for a internal talk. I found a photograph with a zoom of Lena's eye with a legend like "JPEG2K 1% vs JPG 1%", but the webpage was in Japanese, and I can’t reed it. So I use Google Translate to be sure that it was a comparation of the two methods with the same compression level.
So Google Translate is very good to get an idea of what some web page means, but it is not good enough to make a final version of the translation.
This app have some additional difficulties that GT doesn't need to solve: they need to OCR the text in the wild, they have less computational power and they have to do it in real time. It is almost incredible that they can solve these things. With a word to word translation, I think that you can get a good enough translation 90%, 95% or even 99% of the times, but the corner cases can be really unintelligible. The translation of "Tongue Bolivian ..." is fine, and the user can understand what it means, so it is useful. The translation of "Dead End" is something that they should improve in next version.
I once saw this forum post that was in German, that Google translated "A: Nyet." to "A: Yes."
Since it was a question and answer post regarding information for an upcoming game, there was quite of bit of fuss over it before this translation error was discovered.
Some times, but not always. Not even sure if it's "most" of the times. "Right turn" to "correcto turno" is so wrong, anyone following directions with that would be lost pretty easily.
Still, amazing technology, specially since it's all client-side. Hopefully they'll have some contextual analysis in the future or enable Google Translation API use.
John, outstanding work. Word-for-word is IMO more interesting too, because you understand enough to get the point across - but you still get the flavour of the language and enjoy the idioms. Huge, huge potential. You win the Internet today.
Thanks for the positive feedback (I'm one of the developers). We never imagined we'd win the internets!!! :P
About the word for word thing, that is mostly correct to say it's word-for-word. We have just a few short phrases in there, like "por favor". If that didn't translate correctly, it would have been teh lame.
That's great too. Text detection + character recognition + 3d position estimation + translation, each one of the problems is difficult but state-of-art method usually works quite well. The combination, however, is super hard for a small team because each one of them requires special knowledge in that particular manner. To pick and choose which one to focus on and can still get awesome result is a rare skill.
I have some experience working with machine translation during my thesis(using Latent Semantic indexing, LSI), and after watching the video I was very impress with the speed of the translation, specially knowing how resource demanding are other strategies like probabilistic translation or LSI.
Are you hiring? ;) I am a native Spanish speaker btw.
It becomes "make correct shift" (as in "I work in the night shift"). The word in Spanish would be "giro". The natural translation would be "gira a la derecha".
"make right turn" is apparently hard! Word Lens translates it to "make correct turn" but where "turn" is in the sense of taking turns not in the sense of turning. Google Translate gives "to the right which in turn".
297 comments
[ 3.4 ms ] story [ 175 ms ] threadRaise the price. For the love of god. Raise the price. This is much, much more valuable than five bucks. There have been times in life where I'd gladly have paid a dollar per minute for this. And I'm a cheapskate.
Absolutely, read http://bit.ly/bdHKmm to understand why. (Sorry for the shortened URL, the URL recognition software here doesn't understand URLs with apostrophes in here, so it actually is necessary.)
Something like this should not be $5.
We have a lot more Vietnam refugees here than we do Chinese folks. If you'll notice, the three main languages you can get government papers in (at least over in San Jose, no experience elsewhere) is English, Spanish, and Vietnamese.
Incidentally, there's other and easier untapped startup opportunities for the Asian language demographic, if anyone wants to send me a note via gmail
My in-laws are a mix so either language is good, both is better. I can promise 3 Android and 1 iPhone sale just in my home :)
The dictionary is free and the OCR is available as an in app purchase for $15.
I'm just saying this because it is not trivial to adapt an OCR engine to non-latin scripts, because the image analysis techniques are rather different.
I understand that Chinese is harder, for many reasons. It is also much harder for humans. But that is why I'd happily pay, say, $10 per day of my trip for an engine that gives reasonable hints. Maybe more. Try me and see.
Of course, once this technology is ubiquitous and the price of such a thing has fallen to nigh-zero it is going to change the world. But we have to walk before we run, and you'll need the money for R&D and legal costs.
The awesome thing is that once the platform is there, they could easily open it up to other languages and get them rolling too--though they might have to really think before diving into languages with a different character set, such as Japanese.
From what I understand, this is the only app that does that.
I've known about this application for a long time, Otavio showed me an early prototype of this running on his laptop over a year ago. I'm ecstatic that they finally released it.
(I'm also happy that I was able to get their story on HN before TechCrunch :D)
And I’m sure that it will also, by effectively removing all barriers to communication between different cultures and races, cause more and bloodier wars than anything else in the history of creation.
See what you've done?!
i hope this app will be nice and disruptive, and we'll be looking carefully at what people expect and how they actually use it. it's a platform, and we're really excited about the directions it opens up.
thanks for the link, cheers!
Agree I can't read it, pretty simple request.
Aside from that, and as a Spanish native speaker, the translated text sounds really weird to me. Even without knowing English I would be able to understand the translated text (at least the example that you gave) but it can be improved. But this is the only con that I see to your application. Probably you're already working on that, but I thought that you might like this input.
Great idea!!
This could just be a case of needing to optimize and train their recognizer to deal with a much larger set of possible characters, or it could require implementing kanji-specific OCR techniques like attempting to decompose the characters into their constituent strokes, and recognize based on classification of those strokes (orientation, position, direction).
On the other hand, an English to Chinese translator might be more doable, and might also be more commercially profitable.
An app that converted from characters to pinyin (a much easier problem) would be gold to someone trying to learn mandarin. I'd easily pay $50 for something with that functionality without it even making an attempt at the english. It's a much smaller market than those trying to understand chinese signs on a vacation, but it's one with a much larger stake and interest in the result.
Here's a video demo: http://www.youtube.com/watch?v=x7VTo0656Rc
Of course, guessing the correct word when performing word-for-word translation of hanzi is almost impossible, so even the extremely primitive product I'm thinking of is very difficult.
With such a product, you should perhaps brand yourself more on your product than your company name. :)
Just throwing it out there.
In this case someone has already registered wordlensapp.com. We can only hope it's the Quest Visual people.
Different registrar, different DNS and the obvious spam landing page. sigh
There are many trends for using domains similar to what you would desire: get<product> being the best, and <product>app being one of the other choices.
I have very bad experiences with finding an app's site, if it models <product>app. Maybe it's just bad SEO on their part, but I imagine that it'd be an awful hassle for any normal user who don't know the conventions.
Then it struck me that <use>product would be a very interesting domain that makes much more sense than <product>app. It also mirrors the imperative <get>, although it may not be as popular and known to users - yet.
I think it's a shame they didn't secure a regular <product>.com domain for such a great product - but, on the other hand, I'm sure all the press will forgive them (hell, they're trending on Twitter, and 80% of my social media digest today was about the bloody thing).
I've registered usewordlens.com, usewordlens.net, and usewordlens.org and will be happy to hand it over or point it to a domain of their choice (if someone tells me how the hell to do it using name.com, because I'm a complete idiot in that regard. I guess I have to mess with some DNS).
otaviogood or johndeweese, if you're reading this, my email is in my profile.
http://techcrunch.com/2010/12/16/world-lens-translates-words...
Or anyone for that matter. Make a business, be king of the world. You deserve it.
The first card says "welcome to the future" which I thought was perfect :)
Would like to see what it really does.
What is the ETA on French and German?
I was using a 4th gen iPod Touch, so lower-res camera and no flash.
Unfortunately, I was not able to get it to work as well as I'd hoped. I tested physical copies of a few things:
1. Mac OS X Snow Leopard cd case. Black sans-serif text on white background. This worked the best, with the word-reversal consistently getting "Snow" and "Leopard" reversed with little flicker. "Mac" seemed to go in and out. OS was rarely replaced, and usually along with "Mac" as though it were part of the same word.
2. C++ Programming language book cover (http://pixhost.info/pictures/631454). Dark blue serif text on white. This almost never worked. When it did recognize letters, the recognition shifted so much that the word was a constantly moving jumble.
3. Throat Coat tea bag. White serif text on Red. At any point in time, about 50% of the words were recognized and reversed.
You can take a snapshot, and each word it recognizes is highlighted, which is pretty cool.
I would definitely buy something like this that handled non-romance languages to English.
I believe the iPhone 3GS/4's autofocus makes a huge difference in clarity for close-ups.
I admit I was just thinking about reading foreign language books and translating some parts, this app takes it much further.
At least in my limited experience, deciding what to work on is really important.
Acknowledging great ideas doesn't take away anything from their superb execution.
I did: "One that I've mentioned in almost every post here for ages, a service that I can take a photo, it does OCR and feeds the text into Google translate. Bonus points if it does source-language-detection. More bonus points if it is actually usable in a foreign country on real live things like signposts, menus, timetables, advertisements, book covers, leaflets, etc." - me, http://news.ycombinator.com/item?id=569838
Nearly two years ago, travelling in a foreign country with an iPhone, knowing Google Translate existed, knowing OCR existed, and not being able to do anything with that knowledge.
(It's so rare that I hit this kind of advance view on things I'm going to be shamelessly happy about it. NB, I'm not claiming anything here - I couldn't write it and didn't try).
The creators will likely see a payoff from this whether or not they do any of these things correctly. But the magnitude of that payoff is entirely dependent on those ifs and the difference between executing correctly and not is humongous.
Seeing this thing in action in real life is life changing.
I had to put the App to test.
Printed some simple English words and tried it out.
Original paper:
http://imgur.com/EYsko
What the App shows in REALTIME:
http://imgur.com/FhaSW
Comparison with Google Translate:
http://imgur.com/GOd94.jpg
[EDIT: Note, Translating Muerto Fin to English in Google Translate, does result in 'Dead End'. Can any Spanish reader clarify why the original Google Translate chose to translate 'DEAD END' to 'callejón sin sali'.]
[EDIT: So let me get this straight. Their program, running on iPhone [256MB RAM, 600 MHz ARM CPU], can take a live image, perform OCR, translate, create another image with the translated words. And all of this happens real time? Wow.]
I think it's really hard to get a 'Wow' reaction from HN crowd. And you have hit a home run! As someone else said in this thread, you are going to make boatload of money. Congrats!
On the other hand, "muerto fin" has NO meaning in spanish, (and that's what you get when you translate word for word without context).
E.g.
[Search Term] -> [Interpretation]
New York -> New York city
New York Times -> The news paper
New York Times Square -> Famous tourist spot in NYC.
There were other examples in the article. I will try to find that article and link it up.
I think Google should buy these guys and provide them with their knowledge of 'context'. Google Translate team and these guys should talk right away.
The best examples of this, I thought, were the following passages:
Yesterday I went to the symphony. The sound was beautiful.
versus
Yesterday I went to Long Island. The sound was beautiful.
In the first passage, sound refers to an auditory experience; in the second, it refers to a body of water. Within the sentence "The sound was beautiful", there's no way to know which to use, so one must look to surrounding information to make a guess.
But I've given a short and clear example; anyone can see the right solution. But in real-life usage it's frequently unclear.
Yep... solving "context" 100% is equivalent to solving A.I.!
First: This app is amazing!
But I think that this app is doing almost a word by word translations, without a list of usual expressions or grammatical analysis.
For example, the correct translation of 'Dead End' is 'callejón sin salida', that means literally something like 'street without exit'. Translating word by word 'Dead End' you obtain 'Muerto Fin' that is unintelligible.
Another example from the video is the translation of
The app translates this as But the correct translations is something like because in a translation between English and Spanish you must reverse the order of the adjectives and nouns.Real time OCR + word for word translation....maybe its not particularly technically interesting on paper, but to watch it on a handheld device in front of your eyes is breathtaking.
Do you have a timeline for other languages? (Japanese)
But I think that the real problem are the idioms and phrases, like "dead end", that have a completely different translation.
On the other hand, I think that you have more processing time that 1/10 sec. Most of the time the user will point the app to the same text for 10 or 15 seconds. I think that it is possible to show first a very fast translation almost instantly and a few seconds later show an improved version.
For example, a few years ago, some students of my wife give to her a homework about clocks and gears. When she read it, she was annoyed because the redaction was incredible horrible. But later we realized that the students didn't write it, the "homework" was a web page translated with Google Translate.
Another time, I need an example of the differences between JPEG and JPEG2K for a internal talk. I found a photograph with a zoom of Lena's eye with a legend like "JPEG2K 1% vs JPG 1%", but the webpage was in Japanese, and I can’t reed it. So I use Google Translate to be sure that it was a comparation of the two methods with the same compression level.
So Google Translate is very good to get an idea of what some web page means, but it is not good enough to make a final version of the translation.
This app have some additional difficulties that GT doesn't need to solve: they need to OCR the text in the wild, they have less computational power and they have to do it in real time. It is almost incredible that they can solve these things. With a word to word translation, I think that you can get a good enough translation 90%, 95% or even 99% of the times, but the corner cases can be really unintelligible. The translation of "Tongue Bolivian ..." is fine, and the user can understand what it means, so it is useful. The translation of "Dead End" is something that they should improve in next version.
I once saw this forum post that was in German, that Google translated "A: Nyet." to "A: Yes."
Since it was a question and answer post regarding information for an upcoming game, there was quite of bit of fuss over it before this translation error was discovered.
http://dl.dropbox.com/u/5815494/GW-Deutch.jpg
I cannot understand what heuristic would translate the complete German sentence "Nein." into "Yes.".
Still, amazing technology, specially since it's all client-side. Hopefully they'll have some contextual analysis in the future or enable Google Translation API use.
About the word for word thing, that is mostly correct to say it's word-for-word. We have just a few short phrases in there, like "por favor". If that didn't translate correctly, it would have been teh lame.
Oh please, I mean, it's only one of a kind app that millions could make use of :)
Frickin awesome app tho'
Are you hiring? ;) I am a native Spanish speaker btw.
Do you have an API or something? I would love to work on a tamil to english dictionary.
Can you please email me? My email is in my profile. Thanks.
I like how "Make right turn" becomes "Make correct turn".