Taking SRSing to the next level

[HerrPetersen]

In this thread I just want to introduce a programm a member of the Reviewing the Kanji-site created. I guess it is still not fully functional but I am really excited about where it will go. Basicly it works like this:

1. Find a movie and a subtitle file {and an optional english translation file matching it}
   
2. Using the programm will produce a spread sheet and the following data for each line of dialogue in the sub-file:
   

• The hanzi for the sentences of the entries
  
• A sound file ripped according to the timing of the sub-file
  
• A snapshot of the movie at the given time
  
• An English translation (if you have a second subtitle-file)
  

This makes it easy to create an anki/(perhaps)mnemosyne-file.

Here is a link to the original post

Edited February 10, 2009 at 03:48 PM by HerrPetersen

[HerrPetersen]

Here a quick howto and some sample decks:

1. Download and install anki
   
2. Download something from
   
   1. 
      
   2. 宝葫芦的秘密(Secret of the Magic Gourd - a children's movie) - 37.55MB, ??? cards, the deck has been made by wenhailin.
      
   3. Ipman - 14MB, 240 or so cards, all cards are "cleaned up", difficulty varies from basic stuff to (?)
      
   4. fearless - 19MB, 500+cards, My first attempt, quality of subs not so great, quality of rip not so great, first 100 cards have been somewhat cleaned up.
      

[*]Unrar/Unzip

[*]Open anki-file with anki (don't seperate the anki-file from the anki media directory)

[*]Have fun!

I put pictures and the sound in the question field, the hanzi in the answer field. Please give it a try.

EDIT: To create an anki-deck of YOUR favourite movie just go to the LINK OF THE NEWEST VERSION of the program and download here.

Edited February 8, 2009 at 11:47 AM by HerrPetersen

[greenleaf1348]

I tried your deck; pretty neat! I'd still end up deleting 80% of the cards, but that's a lot easier to creating a deck manually.

The main issue is that most movies I acquire have hard subs

[HerrPetersen]

Yeah you still have to delete a lot of stuff, but this is done pretty quickly. Maybe there is a way to cut out the bottom part of a movie so the subs won't show. Subtitles are easily obtained via shooter.cn - however quality is not always so great.

Right now I still have a text-book I want to finish, but after that I am looking forward to go watch a cool Chinese movie, and completly learn all the dialogue via this tool. Just gotta find a very cool movie.

It would be nice to have a database/a list of torrent-files and matching subtitle files.

[greenleaf1348]

well, timing mismatch between subs et movie is my problem; I don't really know how to fix that.

Out of all the mandarin language movies I'v seen, I think "Eat, Drink, Man, Woman" (by Ang Lee) is the best suited to being studied.

[HerrPetersen]

It is probably better to just find matching subtitles. But you can use aegisub, which has a shift-timing option (timing --> shift times).

Good tip, looks like a good movie - will try to find it.

[yersi]

Thanks for sharing that deck, sadly there are a lot of cards where the picture doesn't show up or the sound is cut off. While I don't think it's feasible to learn the dialogue of a movie with this (flashcards usually work best with very short bits of information), these flashcards at least seem to help your listening comprehension by giving you a short line of dialogue, making you guess at the meaning and then showing you the characters.

Since I'm out of new cards for my German deck, I'll incorporate this into my daily anki regimen and see how it works out.

[HerrPetersen]

Agreed, for building up vocabulary this is not so cool, but for listening (and maybe shadowing) this is pretty good practice.

Do you use the option "show new cards in order"? The first cards should be pretty cleaned up. Anyhow that was my first attempt at using the subs2srs-tool, so there is much to improve. In the new version of the tool there is a "pan" option which allows you to cut of the dialogue at -xxx at the beginning and +xxx ms at the end of the recording - so this already some improvement. (I only checked the first 100 cards in the fearless deck, but for those pictures were always shown... strange)

Some hanzi are messed up in the Fearless-deck and there are a lot of mistakes in the hanzi.

I made an "Ipman"-deck just now, which I feel is already better in most aspects.

Also I am obtaining [饮食男女] (thanks for suggestion greenleaf) for which I already found some subs.

Edited February 4, 2009 at 11:52 PM by HerrPetersen

[Radial]

looks like a great idea and I would really like to try it... however, the link to the software is blocked (mediafire.com) and I cannot download. Part of the frustration of living in China.

[HerrPetersen]

@Radial, postmessage me, so I can send the deck via email.

I made a second attempt with the tool and made an "Ipman"-deck. You can get it in my second post. I feel it already came out it little better than "Fearless" however I still want to improve stuff a little before I want to really use it to learn stuff.

Also feel free to try out the tool yourself: subs2srs.

EDIT: misread your post, I will gladly email the prog to you, it is not so big (less then 1MB)

Edited February 5, 2009 at 11:05 PM by HerrPetersen

[Gleaves]

I tried out the Fearless deck last night. Very cool notion, with plenty of promise. Thank you for giving this a shot and making them available. I'll try out the Ip man deck as well.

How long did the Fearless deck take to make?

[HerrPetersen]

Like I said in the first post, after sorting out a problem (which has been solved with the current version of subs2srs) it took about 10 minutes of my time, and the computer around 40 for ripping the snapshots etc.

From the author of the program:

Approximate times for each step of the Toki wo Kakeru Shoujo deck:

1) Demux audio and subtitles from the .MKV file: 2 min

2) Convert AC3 audio to Anki-friendly MP3: 5 min

3) Create folder structure: 2 min

4) Create necessary output formats and run subs2srs: 5 min

5) Split audio track: 10 min (I think)

6) Generate snapshots from video: 25 min

7) Create new Anki deck, move files, and import data: 5 min

[yersi]

I think he said in a post above that it took 5 minutes to set it up and 30 minutes for the computer to process it. Not bad.

Herr Petersen, the Ip Man-deck is great, did you clean it up before you uploaded it?

[HerrPetersen]

Glad to hear you like it, I went through it once and deleted stuff that was messed up or did not seem useful. Going through the movie was not "work time" for me since there were a lot of times where I could read most of the hanzi but could not understand what was spoken - so hopefully some neural connections were made.

It is only CD1 of two CDs and somehow the timing in the subs changed throughout the movie - so I had to delete all the cards after maybe 3/4 into the movie. (Maybe the subs were for another version of the movie?)

[wenhailin]

Herr, firstly I want to thank you for your instructions in one of the other threads, I was having a lot of trouble getting subs2srs to work properly. Anyway, now I have so I will share with you.

This deck is from the kids movie 宝葫芦的秘密 (Secret of the Magic Gourd). Tell me what you think of it.

Here it is

[HerrPetersen]

Great stuff, will begin downloading shortly! Please keep in mind, that the programm is still being improved on an almost weekly basis. By now it reached already version number 6.

The LINK TO THE NEWEST VERSION should always be here.

Edit1: I don't know, if you have known the mediafire page before, but I got introduced to it via the downloadlink of the prog - great place to store stuff!

Edit2: I will put your deck in the second post, so people will have something to chose from - if you don't like it, please scream!

Edited February 10, 2009 at 03:21 PM by HerrPetersen

[greenleaf1348]

thanks everyone for the decks

[HerrPetersen]

@wenhailin - my anki-version does not support filenames with hanzi in them so I had to rename the deck and the media directory to something like "magic" and "magic.media".

Anyhow - the deck looks pretty good and the movie is definitly good for learning everyday-vocab, however there are still quiet a lot of entries, where the timing is off. I have requested a "fine-tuning"-version of subs2srs so you can adjust timing for individual cards - however there is no guarantee that this will be done.

Btw - did you find two sub-files (one English, one Chinese) or did you just translate stuff yourself?

Edited February 8, 2009 at 12:34 PM by HerrPetersen

[wenhailin]

I need to go back through and re-do it all, I realised I made a mistake with the frames per second option which means the images get further out of time as you go.

The subtitles were easily found online, I'm too lazy to do them myself!

[anon6969]

Mediafire is currently blocked in China - is there somewhere else you can host it where I can get decent access to it?