RECENT TOPICS » View all
So yeah, anyone try this yet? I was looking at it as a possible way to enter sentences into Anki without so much pain and frustration from typing. It runs ~$200 (US) for the Asian stripped-down version, but if it saved me time putting things into Anki for me, then so much the better.
I've got a lot of books full of good Japanese sentences lying around, and this might make it easier to... err... not mine, because nobody likes that word... pluck? Yeah, pluck the nice fat juicy ones I want to learn... without killing my wrists or my motivation.
link: http://www.irislink.com/c2-1099-189/IRI … asian.aspx
Wow, never ever heard of anything like that: that's pretty incredible, if the accuracy rate works as well as advertised.
Too bad it's so pricey: it would be incredibly useful for just harvesting books.
Yeah... I know it's pricey, but it looks like it would do wonderful things. Now whether it will *actually* do those wonderful things is a whole other discussion, hence my request for feedback. Apparently it will do vertical column text on the PC (not so sure about the Mac), and the bits and pieces I've found have said that it does a good job of OCR if you get the Asian version. I wouldn't worry too much about the translating version-- I don't think it does Asian languages, anyway. (And having it translate for you would defeat the whole purpose of self-study, right?)
It's not especially cheap, but I'd be willing to drop $200 if it does what it says it does.
Someone needs to take one for the team and let us know how accurate it works. ![]()
I found a review video but it's in Chinese. Maybe someone here who understands it can let us know the verdict. He tests Japanese around 4:30 and it looks pretty good. The mac version of the software looks pretty slick. It seems to have some Spotlight integration but I'm not sure what he's doing at the end of the video.
http://youtube.com/watch?v=9FMOQWci83o
If you have a normal scanner you can do the same with office's language tools.
Transtic wrote:
If you have a normal scanner you can do the same with office's language tools.
That's true: my scanner automatically tries to OCR my drawings whenever I scan them in. I've never thought about applying that to sentence mining (er, plucking) though. Seems like too much of a hassle with big textbooks though.
Well, it looks encouraging, but since I don't know any Chinese at all, I have no idea what he's actually saying.
I checked at Amazon.com, but the only reviews were for the Western language versions, and they were a mixed bag. Then again, this particular scanner has been tuned specifically for Asian languages, so the reviews may not be germane. But I'm not sure if it's the hardware or the software that's been tuned, or just the marketing. ![]()
I found this review at SWET, but I have no idea how old it is.
http://www.swet.jp/index.php/weblog/com … _software/
The reviewer seemed satisfied with it. The idea of using it to cherry-pick just the sentences I want is really tempting, as opposed to having to use my flatbed scanner and find some OCR s/w to work with it.
Just give me 12-24 hours, and I'll probably wind up talking myself into buying one. ![]()
Oh and for the record, I just stumbled across this today in a catalog I got in the mail. I was looking for something PDA-ish (no luck) and I found this thing instead. When I went to the company website, I found out about the Asian edition, and put 2 and 2 together. Whether I get 4 or 37 is a whole other thing.
EDIT: Found a recent review of the product, but it's not the Asian version.
http://www.biosmagazine.co.uk/rev.php?id=669
I also found a link to the product manual:
ftp://ftp02.irislink.com/support/oa/opt … al-eng.pdf
Last edited by rich_f (2008 January 04, 12:20 am)
My Chinese isn't really that great, but what I could make out was quite positive. He was mentioning the relative dpis of previous models (300, 700, etc). He was then saying that the scanner could handle traditional (I believe dude is Taiwanese) and simplified Chinese texts as well as Japanese. He had no trouble with any of them, it was all accurate.
There seemed to be a bit of a snag using Windows and vertical text. I'm not quite sure what the issue was, and I'm not about to go play the guess-what-the-word-is-and-try-to-look-it-up game that I used to play in the PRC, but he did tell us to take note of it.
He seemed overall pretty happy with it. I wasn't sure what the point of the Spotlight thing at the end was, though. Please take what I said with a truckload of salt as I am not a Chinese expert. Someone with more Chinese skill (native speakers, anyone?) should, if inclined, should definitely add to / refute what I'm saying to clarify.
From what I'm gathering on the review, though, this looks good for sentence...uh...mining - unless you worry that hand-copying things actually is helpful (debatable, but a tiny worry in my head). For $200, though - considering that a lot of textbooks (this coming from a recent college grad) can run you $100+ - that ain't so bad. I could probably be conned into buying this, and I'm poor.
Megaqwerty wrote:
Transtic wrote:
If you have a normal scanner you can do the same with office's language tools.
That's true: my scanner automatically tries to OCR my drawings whenever I scan them in. I've never thought about applying that to sentence mining (er, plucking) though. Seems like too much of a hassle with big textbooks though.
You don't need to scan a whole book at once. Pick a short story (a couple of pages), then you can do the following with it:
1.- Scan and OCR the original text
2.- Check it "manually"
3.- Put it onto the "Reading tutor" and make your own glossary (J-J, of course) automatically
4.- Use a TTS engine and record the text into a mp3 file
5.- Print a copy of the text including the glossary, and study it along with the audio version
6.- Pick up sentences and vocabulary for your SRS
Of course, you can add 1 or 2 more steps (looking for grammar explanations, etc)
Whoa, I think I just HAVE to get a Japanese OCR RIGHT NOW! ![]()
Since it needs to be attached to at least a laptop, I don't think it will be that portable. It would be far better if it could store data internally so you could carry it about with you and then download the contents to your PC later. That would give it a significant advantage over a normal scanner.
Hmm, with that in mind, another option might be something like this
http://www.pcpro.co.uk/reviews/110716/p … rc800.html
combined with some sort of OCR software. I have no experience with OCR software so I think I'd test that first before splashing out on a portable scanner. Has anyone used OCR on Japanese documents? Experiences?
Transtic wrote:
Put it onto the "Reading tutor" and make your own glossary (J-J, of course) automatically
Whoa, never seen that before: very cool!
I bit the bullet and just ordered the IRIS Executive Asian from the company's website. We'll see how well it works in a few days.
rich_f wrote:
I bit the bullet and just ordered the IRIS Executive Asian from the company's website. We'll see how well it works in a few days.
Excellent!
I have IRIS standard Asian OCR software, which is very good.
But I will be very interested to see how the pen works. If it works well I think I'll get one. So, please let us know what you think of it...
Mark
Summary-- shaky start, nice finish.
Okay, I ordered the pen on Monday, and it showed up today (Wednesday). For starters, I got an email confirming my order, and then nothing. No shipping info, nada, nichts, rien, 何もありませんでした。 It just randomly showed up this afternoon.
It showed up in what I would call "weird" condition. The shipping box was fine, but it looks as if this is a returned product, or at least an opened one. The product box was slightly torn, the plastic on the software container was torn and half-hanging off, and the inside of the box was torn as well. Also, the CD contained in the CD box was covered in fingerprints. The Asian CD was not.
There is a "Quick Start Guide" in about 6 or 7 languages, but there's no info on the Asian version. The only way you know this is the Asian version of the executive is that the box has a sticker with "Asian Version" on it, and there's a CD with ASIAN on the back of the envelope. That's pretty much it.
The pen itself is small, and plastic. It has a 2-sided button on the side-- sort of like 2 buttons melted into one piece of plastic. The USB cable is 6 feet (a little less than 2 meters) long.
So why did they include the non-Asian version of the software? Beats me. I installed it, thinking that maybe the Asian CD would just be support files... but no. As soon as I popped in the Asian CD, it told me to uninstall the other one. *Sigh* Ok, so they don't do documentation/install very well, huh?
After uninstalling, I installed the Asian version. There were about 10 errors on install, where the installer couldn't find various English and French support files, but I just hit ignore, since most were PDFs. (And I already downloaded the PDF version of the English manual from the website.)
So yeah, it's not a very auspicious start. So let's see how the product actually works.
先生は刺身が食べられる。
Okay, that's straight out of the Dictionary of Basic Japanese Grammar.
Not bad. Only took 3 tries, and that was my first 30 seconds with the product.
危り窃れ?
iり切れ?
定り切れ?
hsjaれ?
売り切れ?
Okay, that was 5 tries for 売り切れ? out of Catchy Japanes Phrases. You need to make sure the scanner has enough "starting room" before scanning. It also managed to ignore the furigana on the 2 kanji, which is nice. Also, this sentence was in a different font-- kind of stylized.
との村では、見事に実った麦穂が風に揺られることを狼が走毛という。
この村では、見事に実った麦穂が風に揺られるニとを狼が走るという。
Okay, that was 2 tries to get the first line from "Wolf and Spice," a light novel. The text was vertical in this case. You need to go into the software, Options menu, Scan Vertical Text.
There's a window that shows up in the lower left corner of the screen that shows a see-through box to see exactly what the scanner picks up. Once I started looking at that and paying more attention, my scans improved.
Also, setting the scanner to small text mode seems to help a little. Maybe?
So for preliminaries, I have to admit that I'm pleased with it in spite of the dodgy packaging. All in all, it looks like it'll be a hell of a lot faster than typing, and I can leave my input method as English. I'm just using it to paste the JP text wherever the cursor is in whatever application I'm using. So it just scanned straight into the quick reply box of the message board in FireFox.
Not bad.
Last edited by rich_f (2008 January 09, 4:47 pm)
After further fiddling, you can move and resize the scan preview window anywhere on the screen you want. But I'd suggest cleaning up your desktop.
Also, the executive will do multi-line scanning, but after trying it in both JP and Eng, I wouldn't recommend it. It's tricky to get a good scanning subject that way, and it's just faster to do it one line at a time.
EDIT: Actually, using the small-text option is a situational thing. Some had suggested to do it to improve overall scan accuracy, but sometimes it would OCR the furigana to replace the kanji, especially when the furigana is below the kanji. Also, to improve accuracy, you need to calibrate it for each book/item you scan. I found that significantly lowered mis-reads. And don't move the pen too fast. A slow, steady pace will give you a better scan.
Understanding Basic Japanese Grammar puts boxes around some text. (Not much of it, though.) The scanner hates this. Not much you can do about that.
Finally, the software is a little on the buggy side. It's already hard-crashed on me once when switching from Firefox to Word. Yanking the pen out and putting it back in, then restarting the S/W seemed to work fine.
Last edited by rich_f (2008 January 09, 4:44 pm)
Thanks for the report rich_f.
Knowing what you know now, would you have bought it?
I'm on the verge of buying one...
suffah wrote:
Knowing what you know now, would you have bought it?
Well, it's a little too soon to say for sure, but yeah, I probably would. I'm still having a tough time justifying the extra $100 for the Executive. I never could find out if there's any difference in models, hardware-wise. (I suspect not.)
There is a difference in software, but I don't know if it's $100 worth of difference, based on what I plan on using it for, and based on how the software itself performs. Someone needs to buy the $200 model and see how well it scans the same sentences in the same books. (I have a crapton of Japanese textbooks and reference books, so that shouldn't be a problem.)
The hardware performs pretty well. The software is a little glitchy, but I haven't checked for a patch yet. The company's customer support doesn't do much to inspire me, and the product I received was obviously an open box model, sold for the new price. (Which kind of irks me. Not that it matters a whole lot in the actual functioning, but when I buy something, I expect new unless stated otherwise. And if it's open box, I expect a discount.)
I haven't had time to mess with Anki yet, but I have messed with it in Word 2007, and it works in that, too. If you have problems with it, try setting Windows' input mode to JP.
Bottom line: It sure beats the hell out of typing, and that pretty much trumps everything else.
It only works with windows? is that right?
It sounds pretty cool, though I would be a bit ill if they sent me a used product. It might be worth requesting a discount, its never to late
.
It works with Mac, too, but I have no idea how well, or what features may be missing on the Mac version.
I know it's pretty difficult to describe accurately but could you try to give us an indication of how quick it is to scan? Are we talking about a swift swipe or a more careful and deliberate tracing?
Is it possible to time how long it takes to scan a relatively long line of text?
It's deliberate, but not too slow. If you go too fast, it doesn't scan well. To scan the first line from "Wolf and Spice," it took about 5-10 seconds per swipe. Then you just hit the button on the side of the pen to hit return (or whatever key/command you want to assign to it) if you want to enter a new line.
So it's about as fast as a fast typist, unless you have problems scanning, in which case it's as fast as a slow typist who makes mistakes.
Since the image you scan shows up on the screen as well as the text it OCRs, you can compare the two pretty quickly. That would probably be the only thing that could slow you down- comparing your scanned text with the actual text.
I'll have a better idea when I spend some time with it later today with Anki. I've got 375 cards to clear in Anki before I start putting in new sentences... ![]()
Honestly the accuracy seems a bit low on the pen scans to me. I am curious if you can use your own OCR software and just scan to tiff format. I use the "lite" version of "読んde!!ココ" that came with the printer/scanner combo that I bought and I would say that out of the 200 or so pages of text that I scanned out of a JLPT2 grammar book, I only had about 20-30 errors total (excluding the annoying habit it has of reading "B」" as "BJ"). In fact I find Japanese OCR about 1000 times less annoying than English. On a side note, I did always select my own text regions. If I let the software auto select the regions then the accuracy would drop quite a bit.
Last edited by dilandau23 (2008 January 10, 5:48 pm)
Scanning with a pen vs. scanning with a flatbed is comparing apples and oranges in a way. Allow me to elaborate a bit:
When you scan with a flatbed, the machine does all the work of moving the scanner head at the proper speed and holding the paper down.
When you scan with a pen, you're doing all of the work. The surfaces are never flat like a flatbed scanner plate, the lighting isn't uniform because of your shadow on the book, the book itself is uneven, and may tend to tip (because you're probably holding the book with one hand, and the scanner with another). And you have to line it up and keep it even, straight, etc. All of these are things produce the sort of resulting scans that a flatbed scanner-based OCR doesn't have to interpret, and probably is not designed to interpret.
So a flatbed OCR is going to be better suited to working on flatbed scans. A pen scanner OCR is going to be better at dealing with pen scans where the scan isn't uniform, even, and precise. Pen-made scans run to the wonky because we're not robots.
So yes, it will make more errors, but that's because you will make more errors than your flatbed will. Thankfully the software is specifically designed with that in mind, so it will do a better job of catching those errors than flatbed OCR will. Flatbed OCR probably isn't written with such a fringe tool in mind, so I seriously doubt it would even be able to understand what is going on, let alone what it's trying to decipher.
It's all about workflow and what you're comfortable with. If you're happy, then stick with it. I like the pen because it's small and doesn't get in my way, and I don't want to mess with scanning whole pages when I only want 4-5 sentences on a page, tops. It really will come down to your personal preferences in the end.

