I think you misunderstood my post : aegisub works great .
It's just that to make any change aegisub must convert your subtitle file to the UTF 8 format as it's aegis standard . It's a conversion , in any case not a corruption .
Unfortunetaly MPC , one of the most widespread mediaplayer out there , has trouble to use this format for MPC softsub fonction . This drawback has even been clearly identified on aegis webpage . MPC is the culprit , not aegisub. And I activated IME japanese but there is no relation whatsoever with this  problem .

Anyway there is no problem anymore as zodiac tipped me off about the playline feature of aegis which is even better to check the result .
Thanks for caring though .

Last edited by ghinzdra (2009 February 20, 11:05 am)

I ve been doing a lot of tests lately with sub2srs and I was wondering how everyone else was doing ?

1st question: on the use of "plain" subs without really editing them through a sub editor
From what I've read it seems obvious that most people are taking the subs the way they are without any manipulation except time shifting , as on the beginning there was much complain about useless lines and time lost on suppressing them one by one .... this issue has been settled with advanced function , time limit ,etc....
Still I keep wondering how is it possible to use the sub this way? I mean I tried twice and there are so many lines that don't make sense if you don't merge them , trim them , shape them through a subeditor like aegisub so that they fit your needs  that I just don't get how can they be used "plain" . Subs are designed to translate a ongoing process , what we are looking for is a single unit of meaning : so not too long lest you aren't able to remember and write/mimick it (I reached the conclusion that as much as possible an extract longer than 10 seconde should be avoided ) . Not too short or you're deprived of the necessary background to get the meaning. And it's a really touchy exercice . So it requires a real analysis.

2nd question : on what is really useless
for some of us it seems also that as long as the sentence is unsterdood  it's useless . Nevertheless I think a well understood sentence can still be worth of putting into your srs . I mean for those who agree with ajatt philosophy it seems clear that if we endeavour to get as much input as possible it's to improve our output . And 90% of the ouput is not some really brillant conversation , rather some meaningless chitchat that we're just copying ....on a first look some worthless 1st grader talk . You don't ponder about the meaning of what you're saying , it just pops up . It's casual interaction.
if I take Deathnote 01 for instance つまねんな　やんでるな　まじで　あんなのみにいたの 勉強してるとこごめね  are easy to understand  but am I THAT confident that I know them so well that I'll be able to USE them automatically ? I think one of the interest of using real japanese stuff  is not only to spark more interest than formal stuff (and so carry on studying complex sentences and rare kanjis without being bored to death)    , it's also to get all the worthless stuff that is actually the really worthy thing . what makes for real fluency . And you need to get that a LOT to achieve real fluency . As far as I'm concerned I take both unknown kanjis based sentences and extremely casual chit chat (as long as I deem that I'm really likely to use it ,that I can envision myself saying that)



3rd question : on the pace
Do you feel you're going through this faster than it used to be when you were maybe putting in grammar book examples into Anki ?

Last edited by ghinzdra (2009 March 08, 7:09 am)

I've been thinking about the different ways of selecting interesting lines and suppressing irrelevant ones.
1st option THE PLAIN METHOD /ANKI
-put the whole stuff into anki and once it's in  suppress them manualy .
advantage :  on the first look you feel you can have more time to actually deal with the file , take it one bit at a time , you do the review and the edit at the same time....
drawbacks : 1 actually it's damn too long : you spend your time on the suppress button and what's more your former review gets in the way of the process
2 it leaves behind unused audio samples in your media folder . As your anki file is meant to stay by you for some years  , I think one should be wary about the media folder size . Take only what is  necesseray
3 it gives you lines the way they're intented to be displayed on the screen as subtitles not the way you need them for anki. As I said before the logic is not the same and if fits the size of your screen and your reading pace  it doesn't fit your shadowing or understanding needs  . You do need to edit them .
I did that for my secund try and I definetely did not like it for the reasons above.
By a side effect ,if you agree that it's ineffective to use subs2srs this way it also questions  the batch function as it's plain stupid to try to gain some time on the processing if you have enough time to edit several sub files (wich is really longer) .

2nd option THE EDIT METHOD/ AEGISUB
- edit the subfile with aegisub  and merge , suppress  and THEN import into anki .
advantages : You deal with most of the previous issues this way as you really take what you need .
drawbacks :  you still have to go through the whole file .
It's my standard method right now .

3rd option : THE POINTING METHOD /SUB2SRS
watch the media (anime, drama , movie ,etc...) and point the time or a specific word  When the show is over , you use advanced function of sub2srs to take only lines in a specific time or including those words .
advantages : funnier (you really watch the media )  and quicker (instead of suppressing  you select)
drawbacks : -it's less accurate than the previous one . You'll have a bit of editing after that .
- it doesn't resolve the format issue : you ll still have some sentences that will be butchered owing to the subtitles logic.
- it entails that what you take is by far smaller than what you drop . If by the end of the show you have something like 70 lines I don't think you're meant to use this method.
I think it's a method for someone at really advanced level and who only needs some very specific line .




About your story method and shadowing mp3 method



I also had some thoughts about your "story method" . After a while I eventually discarded it as it appeared that for one part  every SRS is designed to give different treatment to different difficulty leveled propositions : if you take the whole story at a "4 days-4 weeks -4 years reviews" level then it seems to me  you're going against the very philosophy of the SRS :some propositions/sentences are just harder than others , that's the whole point of the leitner box/srs.  For another part I realized that the deeper your knowledge of japanese  , the scarcer the selected lines . Which means in a way or an another the story is broke up .The figures about useless lines are overestimated in my opinion but they do exist though. And those figures will grow up and up as you get better . How can you really make sense of a story with 5 minutes gap every four lines ?

the "Shadowing mp3 method" makes more sense . But my journey into japenese has learned me a lot of thing among which the value of time . Don't waste your time . There is a lot of great and cool material out there but if it takes too much time to process 99% of the time it's better to take something else that you can use quicker . I definetely could ripp off the audio of the online news for the sentences i'm interested in ....except that I would have to 1 locate the mms address  2 download the video file 3 extract the sound  4 cut the file so that I might get the one sentence I'm interested in .... too much time consuming.
So as this shadowing method requires additionnal manipulation , cut , trim, merge , add blank ,etc... I think I'll go on with my ripp off policy . As khatz I just ripp off  the audio track of the media (which takes me 5 minutes flat , 4 of which are used to something else)  and take it on my ipod . If something catchs my attention and I feel like shadowing , I just have to rewind . For all that I don't think your shadowing method is fatally flawed .

Last edited by ghinzdra (2009 March 08, 5:12 pm)

http://www.snapvine.com/sb/096596c40c3b … 30485c71be

The above link is a sample shadow mp3 I made using iKnow audio clips. I did it using audacity.

Steps:

1. Under projects, import audio (I think 40 files is enough)
2. Goto last file, press f5 (time tool), and "pull" the audio to a time that is twice the time of the audio file above it (ie, if clip above it is 3 seconds, put cursor at 6 seconds). Make it longer if you think more silence is needed for shadowing.
3. Press f1 (select tool)
4. Partially select audio from bottom audio (that you just aligned) and audio above it with your mouse.
5. Under Projects: Quick mix them.
6. Wash wash, repeat till they're all merged.

If anyone has a much quicker way of doing this, I'm all ears.

Last edited by Nukemarine (2009 March 08, 7:13 pm)

Well, I started using subs2srs with "Zettai Kareshi" ep 01. Not quite sure how to describe the way I'm mining it.  It's not about getting a sentence with a new word or phrase, otherwise I'd have stuck with iKnow's Core 6k. A lot of the sentences I know all the words and sort of what the sentence means. It was that 'sort of' that determines if I use the sentence.

Pretty much it's: Look and listen to the sentence to see if it's ok to use. Go in what's good or new about it. Use Kenkyusha e-dictionary for J-E definitions if necessary. Since I know many the words, I'm using the nifty Kojien e-dictionary to put j-j definitions of words. Finally, I write out the sentence (the ONLY time I'll write it out) before giving it a "three" score.

Right now, I'm pretty much using every sentence. As I get used to simple dialogue, then I'll probably become more selective. After zettai Kareshi episode 1, I plan to do Rookies Ep 01, Last Friends ep 02 then Hana Yori Dango ep 01 on the idea this'll spread out the type of dialogue I'll be experiencing (still pretty female heavy though). Reason I'm using the above shows is the subs for them are text based so it's easy to copy/paste/edit sentences.

Way different than the initial thought of just shadowing. Now to see if I stick with it.

Last edited by Nukemarine (2009 May 23, 10:57 pm)

kazelee wrote:

@nukemarine

Last Friends, LOL, that was one of the funniest shows I've ever watched. I know it wasn't intended, but the music that played whenever they were showing that one guy's shoes threw me over the edge.... with laughter. Though it soothed my inner sadist, I got rid of the series after I watched the last episode, やっぱり.

I pretty much mine the same way, cept I write sparsely and don't hesitate the delete cards with simple things like やっぱり, 何で, もういいよ, ええ, and はぁぁぁぁい, on them.

It's like Rookies, it starts off pretty well then heads down hill by episode 4 or 5. However, I did enjoy it so it's a useful show for mining. The dialogue feels real (though I'm a poor judge on that).

Things I'm noting (after only 4 days): takes about an hour to add 20 cards. As I go through them, the plot begins to make sense. Odd considering I've seen it twice and read the script once already. Makes me realize the mind makes connections that are not there in absence of understanding.

Hey Nuke, why do you show the kanji and kana sentences on the question side? It would improve your listening ability if you put them on the answer side.

Taken from the other subs2srs thread:

magamo wrote:

As for Code Geass, one of members from UK had been working on the series and some Japanese members helped him a while ago. I haven't seen him for a couple months, though.

That was me, actually. I had stopped while I got around to watching R2 (so I could look things up without getting it spoiled for me), and lately I've been busy with exams (which end in about 2-3 weeks).

I watched R2 raw, but I have .mkv files (Japanese DVD rip, fansubbed) for the first season. Is there a way to extract the subtitle files from them (then replace the English subs with Japanese)?

Is anyone still interested in making site for downloading premade decks?
Maybe we could have a thread with links to the decks (rapidshare etc). I don't know the legal ins and outs though...

I've uploaded a deck for FF: Advent Children, is it ok to post it here?

What I do is add in word definitions / readings as I go through them, if I already know all the words then I delete the card.

But isn't translating between languages which you don't understand perfectly a problem in the long run? I think I read that on AJATT, but I think he was talking about English->Japanese. Feel free to enlighten me though.

cjswanson1355 wrote:

but it's in the .sub format, and subs2srs will not recognize it.  I have tried some converting programs, and all of them either do not recognize the kanji or simply ruin the encoding. Does anyone know how to fix this problem?

You must OCR this type of subtitles. Check this:

http://www.d-addicts.com/forum/viewtopic.php?t=16017