Hi all,

I'm a university student currently in an entrepreneurship course, and my team is trying to gauge interest in a potential app we hope to build so we can decide if it's worthwhile to make. The app would mostly be geared towards ex-pats or study abroad students, but I'd love feedback from other people too. If there's enough interest, we will make this app and hopefully have it on the market in 3-6 months, i.e. this is far from hypothetical.

The app would basically be used to take a picture of some text you don't understand (street sign, passage in a novel, dialogue in manga, etc.), translate it and look up readings, make a flashcard out of it, and then add it to an Anki deck. As an SRS user myself, making the flashcards was always the biggest hurdle, so I usually resorted to pre-made decks, but those didn't feel quite as relevant to my life. Also, the frustration of adding cards turned me off from reading books in Japanese because of how long it took to add sentences to Anki. Hopefully this app would solve those problems.

So my questions to you are:
0. Who are you (ex-pat, college student, Joe Schmo learning Japanese, etc.)
1. Would you use it?
2. Would you pay for it/how much would you be willing to pay?
3. Are there any must-have features you think should be included?
4. Any other suggestions/comments/advice?

Thanks!

Demo it for first to show us the features. The things you mention are easy to do. The problem is they are done poorly. If your translations are better than google translate (which suck. actually i believe all off the self japanese translation software sucks) that is something you must show your potential clients.

Nothing personal but without any evidence to the contrary your project does not look feasible

Programming image recognition with conversion to text is not 'easy to do'.

As for OP, I wouldn't suggest trying to translate the entire text. There is nothing in existence that does a decent job of properly translating Japanese into English. I would suggest your program breaks the text up into words/particles and displays the dictionary definitions for each. That would be most useful.

Maybe just hyperlink to the Jisho dictionary or something.

In fact, once you have converted the image into a string, just paste that into Jisho and it breaks it up for you. You could then take the elements and put them, along with their definitions, into the flashcards.

eg:

http://beta.jisho.org/search/%E6%AC%A1%E...1%E3%81%9F

Cronos Wrote:Programming image recognition with conversion to text is not 'easy to do'.

It is when you use an premade open source component to do it as these apps tend to use

edit: to add to my original post though, your idea may work with other languages (eg. French) as I believe the off the shelf translation components can do a pretty good job with those

juniperpansy Wrote:

Cronos Wrote:Programming image recognition with conversion to text is not 'easy to do'.

It is when you use an premade open source component to do it as these apps tend to use

The image segmenting and OCR are not impossible to accomplish but the automatic translation is. The program won't be smart enough to know that something is a sign or a page in a book; it can't handle context. Even when there is context for translations, it still falters quite a bit. Just try translating a sentence from Japanese to English in Google Translate.

パワー半導体は、交流を直流に変換したり、周波数を変換したりする際に、電力を制御する装置に使われる -> Power semiconductors, or convert the alternating current to direct current, when you can convert the frequency, are used to a device for controlling power.

There's a reason why translators still have a job.

0. Creative professional working in Japan
1. Yes
2. Yes, 2000 Yen
3. It would just have to be able to make the cards/add them to my deck, no hassle with fields etc. Maybe you could add an optimised sample deck that already has the card format right for import.
4. I'm skeptical about the auto translate feature, if it just had a build in dictionary and a simple field where I could enter the translation, then it would be fine. It would be good to have some sort of on-hold list, for images you don't know the translation yet. These won't be added to anki until (you ask someone the translation and) fill it in.

Good luck