Boosting Anki’s Power With Media Enhancements

Presentation matters. The changes in mechanics aren’t the reason I’d much rather play the Nintendo DS remake of Final Fantasy III than the original, and WoW is much easier to play with a decent graphics card that renders boss effects clearly and gives the fight a smooth frame rate. You can upgrade the interface of your anki deck by adding media to make it a more effective and a fun learning experience. So what exactly can you enhance?

Boosting Ankis Power With Media Enhancements


The biggest thing you can do to make your cards more alive is to give them sound. The internet makes it possible to add audio files to your anki deck even for all those sentences you collected from text sources. The simplest way is through the Google TTS anki plugin, which adds high quality computer generated speech automatically. It is also possible to get some of your sentences read by a native speaker on the site RhinoSpike.

Adding speech can make your anki deck come alive. It is the biggest upgrade to your deck’s presentation you can add, because it engages a whole extra sense, and there are a number of specific benefits.  Even a language like Japanese that has an extremely straighforward spelling system sounds a little different spoken than a simple listing of syllables.

If you have a card with a word that isn’t coming up much in your immersion listening, you will pick up the pitch accent faster if you are hearing the sentence spoken every time you review the card. Adding listening to your anki experience increases your total listening time, which I think is particularly helpful in the early stages when you’re getting a feel for how the language sounds.

1. Adding audio to your deck

The first thing to do is to add an Audio field to your card template.  Do this by clicking the “Card Layout” button in the card browser, add an Audio field in the fields tab of the card layout window, and add it to the card by adding {{Audio}} in the answer section of the card templates tab.

It doesn’t go on the question side because then we wouldn’t be tested on the kanji reading. The next step is to put mp3s into this field on all the cards; there are several ways do do this, one of which works for everything and takes almost no work.

Once you have an audio file to add, you can do so by dragging and dropping it into the Audio field or pressing the blue speaker button or F4.

2. Audio from the source

If your sentences are from an audio source to begin with, you can use the actual audio. Adshap has written about how to turn shows into mp3 files for creating your immersion ipod, but for this particular job my favorite tool is subs2srs which I will write about more in a later article in this series. If subs2srs isn’t an option or doesn’t appeal to you, Audacity is a good and free tool once you install mp3 support. To use it to get a piece out of a long mp3, open the file, zoom in, select the portion you want (you can use the play button to test that you have the right selection) and then export to a new mp3 with “File > Export Selection…”.

I think this is the best kind of audio to have, but I don’t have much of it because most of the sentences I’m adding currently are from manga or dictionary diving.

3. Robot voice with the GoogleTTS plugin

(Update 5/7/2013): For Anki 2, AwesomeTTS is the new plugin that has taken over. You can download the easy to use shared plugin here.

The easiest way to add audio is to download the GoogleTTS plugin which you can install from the “File > Download > Shared Plugin…” menu option.  This automatically accesses Google’s high quality Text-To-Speech software.

You can then easily add audio to all cards that don’t have it by searcing for “-Audio:_” in the card browser, selecting them all (click on one then hit Command-A on a Mac or Ctrl-A on Linux and I assume Windows) and choosing the “GoogleTTS MP3 Mass Generator” option from the Actions menu.

Like the furigana generator, it occasionally reads a kanji wrong.  You can fix this in the card editing interface. Copy the expression, select the contents of the Audio field, hit the speaker button, paste the text, and replace kanji it read wrong with hiragana (or punctuation like “…” that it inexplicably reads out with punctuation it doesn’t like “、”.)

Unfortunately, this plugin doesn’t remember its settings when anki is relaunched. You can fix some of this by choosing the “Settings > Plugins > Open Plugins Folder” menu option and editing the file.  Because it is a code file, you may need to manually tell it to open in a text editor instead of trying to run it.  Then replace “en” with “ja” in the “TTS_language = ‘en'” line. There are also some other options and explanations of using the plugin that you can look at in this file.

I’m pretty happy with the way the resulting Japanese robot living in my anki deck speaks. But robots aren’t the only option.

4. Real voices from RhinoSpike

If you want to hear a native speaker reading your collected sentences, it turns out that the internet makes that possible too at RhinoSpike. The site is very simple to use; you sign up, enter some sentences in Japanese, and record yourself reading some sentences in English in order to get your Japanese sentences moved up in the queue.

Unfortunately, this site doesn’t seem to have gotten the critical mass of people necessary to take off, but it is something to look into if this interests you. If I decided to make use of this site, I would try to make some Japanese speaking friends on it and make recordings for them in exchange for getting my sentences recorded.

While there are occasional times when audio isn’t practical to review with, such as using AnkiMobile in public without headphones, it can be muted at these times and I think it is highly worth adding for the benefit whenever you can listen to it.


While audio may be the biggest upgrade to the learning environment of an anki deck, graphical enhancements are the biggest upgrade to the aesthetic experience, as well as providing significant practical benefit.  So how exactly should you set your visual cards to achieve maximum results?

1. Pictures

Pictures are useful on the answer side for illuminating a definition, but they can also be used on the question side. I’ll admit that I first did this just because I liked how it looked, but I discovered that the context it provides is actually quite valuable. Sure, this sometimes gives hints that can give away the answer, but I think it’s worth it. Connecting the word to the entire situation it’s in makes me feel like I truly understand the word more than I can with just a sentence.

To get a picture from a manga or other book with pictures, the first step is to get it onto the computer. If you have a raw scan on your computer, that’s the easiest way, but you can also do it yourself. I actually prefer a camera to a scanner for this, because it can get the pictures close to the spine without damaging it. It’s important to hold the book flat and the camera directly above the panel so that it isn’t distorted.

The picture may need nothing but cropping or it may need more processing. If it just needs cropping, you can skip some steps by using the sreenshot-a-rectangle-into-the-clipboard feature of the Mac OS and then paste it into an anki card. Mac users can use cmnd-ctrl-shift-4 and there are various solutions on Linux. In Windows you can easily use the snipping tool to crop any screenshot and just copy and paste it directly into anki.

If the manga has furigana you will want to erase that, and if you got it into the computer using a camera without a good white balance you might want to correct the color. I use Gimp for this.

To make it greyscale, use the Desaturate option in the Color menu; set it to “Average”.

To erase furigana, use the eraser tool. You’ll probably want to make the image display larger (typing + zooms in) and the eraser smaller.

If you decide a workflow where you crop in Gimp works well for you, that option is found at “Tools > Transform > Crop”.

Note: An easy option for Windows users when using the snipping tool is to use the pen option, choose the background color of the text (usually whitish), and just write over the furigana as you would with an eraser tool.

To get a picture from a show or game, you can take a screenshot, or for shows I would suggest subs2srs which I will be writing about in the next and last article in this series. If you try to take a screenshot of a show, you may be surprised to find that the movie portion of the screen is a solid color. Some movie players have a screenshot option, and Gimp has a screenshot feature located at “File > Create > Screenshot…” that is not subject to the same problem.

2. More text

Sometimes you don’t have a picture to provide that context, but you do have some more text that can. This simply requires adding two new fields to the model the same way that an audio field was added in Part 2.  Another way these fields are useful is when you want to SRS a phrase out of the middle of a horrendously long sentence you can put the other parts of it in these fields.

3. Colors and fonts

When you’re learning Japanese you’re learning a lot of new complicated symbols and it’s a good idea to not make it any harder than it has to be with a hard to read color scheme. What colors people find easy to read can vary a lot, so fiddle around with the colors until you find what’s easy for you to read rather than taking someone else’s answer. But also pick something you think looks nice, because you’ll be looking at it a lot and you want to enjoy the experience as much as possible.

Fonts matter for about the same reasons. When you go to change the font, there are a ton of fonts in the font list and only a few of them include Japanese characters, and some of those are Chinese fonts that may look nice but aren’t always quite right. Japanese fonts included with operating systems include at least one of the words Gothic, Mincho, Hiragino, Osaka, or Meiro except that Hiragino Sans GB is Chinese.

Mincho fonts are serif, resulting in a more brush-like look, and Gothic fonts are sans-serif, resulting in a more pen-like look, though both are computer styles rather than handwriting or calligraphy styles. The fonts in my screenshots are Hiragino Kaku Gothic ProN for the expression and context fields, and Hiragino Mincho ProN for the definition field; these are the two Japanese fonts included with iOS.

The background color is set in the Card Templates panel of the Card Layout window
(Note: You can also access this menu by clicking on the pencil and pad icon to the left of the magnifying glass.)

(Update 5/7/2013): This post shows how it looks in Anki 1. Anki 2 will look slightly different.

Colors and fonts for each field are set in the Fields panel

4. Putting it all together

The remaining thing is how to organize different items on a card. Personally, I like to have the answer like the question but simply add the furigana and definitions/grammer notes. I have the “Hide question when showing answer” box checked in the Card Layout options and format the answer the way I would like the whole thing to display.  In order to get the spacing right when there are fields like Context that might or might not have content, {{#Field}} and {{/Field}} tags are used around linebreaks; these hide everything between them when Field is empty. Here are my formats:




<br />

Kanji stroke order diagrams

Quick, define stroke order diagram. Stroke Order Diagram (SOD): a picture guide used to learn how to write kanji. Okay you’ve seen that before even in the most basic beginner textbook. But wait, don’t get bored just yet, one more definition for you. Colorful Stroke Order Diagram (CSOD): a colorful picture guide used to learn how to write kanji . . . and be awesome.

As I started to think about this topic, I realized that what I had originally done to get CSODs into my own deck wasn’t what I wanted to tell you to do.  I tried to figure out what could be done to avoid certain problems that I have faced using them and the issues behind CSODs grew into a horrible monster. I was ready to throw out the whole post, but instead I was motivated to do something better: defeat this monster, take its loot, and create a better way to make CSODs.

(Update 5/7/2013): This post outlines how to configure the CSODs for Anki 1. With Anki 2, it is much simpler. You download the easy to use shared plugin here.

1. Why use stroke order diagrams?

Adshap doesn’t like writing. But after looking at these, even he thinks they can provide a nice little (colorful) punch to your RTK mod deck kanji. I don’t think learning perfect stroke order is necessary. But I do think studying it can be helpful; thinking about it is an additional way to engage with the kanji and can make the pieces easier to remember, good stroke order habits can make you more understandable to handwriting recognition software, and it can make your handwriting easier to read if you write sloppily.  Or you can just use them as eye candy on your kanji cards.

2. The diagrams

I’ve written a script to color the strokes and stroke numbers in KanjiVG files (which are free to use and share under the Creative Commons Attribution-Share Alike 3.0 license), resize them, and rename them so that they can be dropped into an anki deck.  I used it to create two sets of SVGs that you can use to very easily add CSODs to all of your kanji cards.

One colors the strokes in rainbow order, which is nice because you can see at a glance how the kanji is put together but has the disadvantage that stokes next to each other are similar colors, and color is sometimes necessary to tell which number goes with which stroke.

The other maximizes the contrast between any set of consecutive strokes.  (This uses the magical golden ratio,  which is math hard at work to make your life better.)

You can click any of those to see the the size these are on cards.

3. How to add them to all of your kanji cards

  • Download the collection of your choice: spectrum or contrast.
  • Unzip it and put the contents in your deck’s .media folder (which is most likely in your Dropbox/Public/Anki folder if you sync media or next to your deck at [My ]Documents/Anki)
  • Add <img src={{text:Kanji}}.svg> to your card template, where Kanji is the name of a field that contains a single kanji character.
  • Close and reopen your deck.
  • If you aren’t going to be adding any more kanji to your deck and you’d like to delete all the extra files, you can run “Tools >> Advanced >> Check Media Database…” and select “scan and delete”. If you do this and want to add more kanji later, you will have to repeat the above steps again.

4. A font to match

Want a font that writes characters the same way CSODs do? Go to this Epson Japan font page, download the file and install the fonts, and use “EPSON 教科書体M” or more likely “EPSON ã≥â»èëëÃÇl” depending on how your computer interpreted the pre-unicode font names (this is in the epkyouka.ttf file, but the other fonts are nice too so I recommend installing them all.) This is a font in the family used for elementary school textbooks because the shape is similar to the calligraphy taught in school. Want this to work in AnkiMobile? See Embedding Fonts.

5. Go Forth

Don’t blame me if your hand starts getting a workout from all this writing bliss. You can even work on your inner monologue and count in Japanese as you write out the strokes. Let me know what you think of these CSODs.  Is there any way you would like me to improve them? If you provide some insightful info, I may change them just for you. Really.


Subs2srs can be used to dump a lot of sentences of dubious value into your deck, but it can also be used as a source to pick some very nice cards from, with everything but your J-J definitions already included.

“Subs2srs”, more S’s than you could possibly ever want in an abbreviation, stands for “Subtitles to Spaced Repetition System (ex. Anki).” This powerful tool gives you the ability to turn any Japanese video into automatically made Anki cards with minimal effort.

While it occasionally gets a bad name because it makes it easy to dump a whole show of sentences into your Anki deck which may have dubious value, it really makes wonderful cards. They include native speaker audio, context, and a picture–everything you could possibly want on the card except J-J definitions (by default it can provide whole-sentence English translations.)

So of course the thing to do is take the good and leave the bad. Generate a deck full of Japanese sentences and media. Take from it only those sentences that are worth learning to you. Leave out the English translations and add your J-J definitions.

This is a bit of a long post, but stay with me to till the end, and you will have an incredibly new and impressive tool at your disposal.

How do you create a deck?

  1. Subs2srs and a windows installation to run it on

Download subs2srs here. There are full usage instructions here.

Unfortunately, subs2srs doesn’t run on a Mac, and while versions through 24.0 include linux support I couldn’t get it to work. But because subs2srs is used to create the deck rather than use it, you just need access to windows for enough time to process your stuff into anki decks.

2. A movie file

Subs2srs supports most movie file formats. You can convert a DVD to a file using the free/open source application Handbrake.

3. A subtitle file

The trickiest part with subs2srs is finding Japanese subtitles.  Here are sources I’ve found:

If you can’t find Japanese subtitles, and you are planning on entering phrases from a video source, you can use English subtitles for the timings. If you are lucky, this will line up reasonably well with the Japanese lines, and you’ll have images and audio to go with the phrases you enter.

4. Create the deck

The subs2srs page I linked above has pretty good instructions so I don’t feel the need to duplicate the information.  Except that making numbered instructional pictures is sort of fun so I’m going to include them anyway.

The one thing I do differently form the default subs2srs settings is the inclusion of context lines, which you can do in the context tab of the advanced subtitle options window.  Other things to be aware of is that it is often necessary to fiddle with the subtitle timing options, which I’ve marked with an asterisk, and take note of the information in the completed window which you’ll need for matching the fields up correctly when you import into anki.

Then create a deck in anki with a template that has all the fields you want; note that each context line is a separate field.  Choose import from the file menu to import the tsv file that subs2srs created, assign the fields as subs2srs created them.  Select all the cards in the browser and run the regenerate readings option in the actions menu.  Lastly, put the media folder in the appropriate place.

5. How to use the deck

I do not recommend reviewing an entire subs2srs deck, since most of the cards in it don’t meet the criteria of containing one or two new things you want to learn. Instead, use the deck as a source for sentences that you can move into your main deck.

  • Find the cards that contain sentences you want to learn. One way to do this do this is to search for things in the subs2srs deck when something you want to learn comes up while watching the episode or movie you ripped it from. My approach to manga or books is to read a chapter without interruption except for possibly quickly marking things the first time through, but to go back through it and stop for looking things up and making cards during subsequent times through; I think this would be a fine approach to an episode as well.
  • But what I really like to do with a fresh subs2srs deck step through an episode line by line using anki’s cram feature in the tools menu–I understand more when I go through things slowly this way.  Subs2srs has tagged everything ShowName_EpisodeNumber so you can use that to choose the whole episode. Select the “in order added” option to start from the beginning, or “since last modified” to start where you left off the last time. Since you’re not reviewing it, just pass everything. Mark (the star icon or ALT-M) the things you want to add to your regular review deck.

  • Export the marked cards into a temporary deck using the export option in the file menu; limit to tag “Marked.”  I use a tab-separated file instead of anki deck because then the cards count as new when I add them to my main deck, which puts them where I want in my new card queue.  Also it avoids an issue with editing models (never let a deck contain different versions of a model with the same name.)
  • Import the temporary deck into your main deck using the import item in the file menu.
  • Copy the media from the subs2srs deck’s media folder to into the media folder for your main deck.  Do this while the deck is closed, to avoid confusing the media database.  (You only need to do this once.  If you are short on space and don’t want to duplicate all the files, you can export to anki deck each time and copy over the media from the temporary deck’s media folder.)
  • Add your J-J definitions (or J-E if you’re still in that phase, but even in that case I recommend the J-E definitions format rather than the translation that sub2srs would provide if given a set of English subtitles.)

Important Anki Media Reminder

If you sync your anki deck with the internet to review on multiple devices, you will need to set it up to store the media in a dropbox account. I won’t go over the details since there is already decent guide to syncing media on the Anki site. Once this is set up it does everything automatically. Also, make sure you already have the Japanese Support plugin for furigana.

Make Your Anki Shine

Hopefully some of these improvements will change the way you interact with one of the most powerful tools in the world.

Related posts:

The following two tabs change content below.


Anki technical guru. Learning Japanese because it's a fun game.


Boosting Anki’s Power With Media Enhancements — 118 Comments

  1. I really like how some of those look. However, I’ve recently taken a liking to doing my reps on my phone while on the bus/waiting for the bus and the only features that ankiweb allows, as far as I am aware anyway, is additional text and colors and fonts — both of which I currently utilize.
    I suppose I could have two decks though: one with media which would be done at home, and one without which I could do anywhere I have a free moment.

    • If your phone supports the regular AnkiWeb interface, which I think looks just fine on a small screen, you can use pictures and possibly mp3s; AnkiWeb supports mp3s with ASCII filenames but that feature isn’t compatible with all browsers. Even if media wouldn’t work on your phone, a lot of media makes reviewing better but doesn’t render the card unreviewable if it doesn’t show up, so you can add it and just experience it when you happen to be doing reviews at home.

      I do a lot of my reviews using the AnkiMobile iPhone app. There’s also an Android app.

  2. Just in the past few days, google slowed down the speaking speed, which I’m not entirely happy about. It sounds like at some point they’ll make this configurable but it isn’t yet. When they do, I’ll post instructions for making the plugin use a more natural speed again. (Let me know if you notice before I do that it has become configurable and the default is still slow.)

    • I’d be happy to:

      Background: 234, 236, 211 (#eaecd3)
      Expression/Reading: 0, 0, 0 (#000000)
      Context: 0, 85, 0 (#005500)
      Explanation: 85, 0, 0 (#550000)

      I wasn’t sure whether decimal rgb or hexcode would be more helpful so I gave both.

  3. Thanks for the tips. I can use this to take screen shots with my iPhone. I may not do this too often because I find it to be a little too time consuming; however, your colors and fonts are FKNスゴイ!

  4. I’m having some trouble getting this to work.

    The actual images in the folder are .svg files so I tried changing the image link to but I still get the missing image picture in Anki. Can Anki open that type of file? Is there a place to download the pictures in .png format?


    • That was a typo on my part! I should have said .svg instead of .png. I’ll ask Adshap to fix that. Sorry for the bad proofreading on my part.

    • Also: If people want I can upload PNGs as well, but SVGs are significantly smaller and should be supported by any anki use except ankiweb in an old browser.

      Thanks for pointing out the problem. And I’m glad to hear you like them.

    • Ack, I’m not reading… usually I sit on comments for a bit to make sure I’m saying correct things. You said you changed the link. Did you copy and paste rather than typing? I see that wordpress changed my quotation marks to styled ones; they’re not actually necessary so it’s probably better to just remove them. And the third thing to try is the media check even though I remember that just being necessary at the syncing point in the process, but I don’t understand all the technical details behind why this works so I’m not certain that should be the case.

      • I’ve tried everything you’ve suggested up to this point, and I’m still unable to get this to work.
        I’m looking forward to being able to try it out on my cards though.
        One thing I’ve noticed is that there are files with names like 繧ア繧・svg instead of 繧ア繧.svg — In addition to it, I don’t have anything in my kanji field that would match up with most of these files either.
        But I’m not quite sure how this is supposed to work, so it may have been done like that intentionally.

          • Ah, my kanji fields are the default ones, so they just contain one character.
            蜈ョ.svg – I just mean that the files are named like this, so it seems like they won’t be picked up by looking for the kanji name then appending .svg to it.
            Also, I just looked in the zip file itself and it has lots of items with the same name, so they overwrite each other when they’re being extracted. Maybe it is an issue of the names getting corrupted in the zip file or perhaps them not working for people on other platforms.
            They’re being read as svg documents when they’re in the zip folder, but then as extensionless files when I extract them as they are named with a ・ instead of a . and my system won’t accept ・s for that purpose.
            This is how it appears for me by the way:
            Hopefully some of this proved useful anyway.

            • I tried it again and it worked! Not sure what was different this time, but it worked with the .svg extension.

            • I used winrar.
              Had to reply here as it seems I can’t reply that deeply.

            • Make sure you have the lastest version. If that doesn’t work, try a different application. If you do get something to work, that could be useful to other people.

            • Ah, I tried a few other programs, and they gave the same issue.
              However, since you created the archive on a mac, I decided to use stuffit, and that worked. So for anyone else experiencing the wrong file name issue, try using stuffit to extract the .zip file.

            • Ah, thanks. I’ll try zipping them up in linux instead. But I’m moving this week so I don’t know when I’ll get to it.

        • Wait… you say the file names are coming out that way? In the unzipped directory? That’s odd. I have tested downloading and unzipping the files and they did come out right (the filename should be a single character followed by .svg except for some variants that will have a dash and the name of the variant after the character (that is because these came in the original data and I chose not to bother to throw them away; they don’t do anything for the automatic anki adding.)

  5. No worries :P

    So the weird thing now is that I’ve gotten the links to work within the deck so the pictures show up on the desktop client, but all of the strokes are behind the background color I have (with the stroke number showing through fine).

    Also, no matter what I do I can’t get the pictures to work on Anki mobile. I’ve checked the media database a bunch of times and checked the database as well but they still won’t come up.

    • Interesting bug with the strokes hiding behind the background. I was able to reproduce it for some card template settings but haven’t figured out the precise pattern yet. Would you mind sharing your card template? And do you have it set to hide the question when showing the answer?

      With the AnkiMobile syncing, are other decks’ media syncing properly? Has dropbox had time to upload the 10,000 files you just dropped in it?

  6. What would be the easiest way to share the template? Right now I have the answer field as only the link to the picture. I don’t have it set to hide the question when showing the answer.

    One thing I was just messing with was that the picture isn’t behind the background if I remove one set of curly brackets around text:Kanji. Will that have any negative effects? Unfortunately it still isn’t working on Anki mobile. The pictures actually all synced to my phone on the last sync but they don’t show up on the cards.

    Thanks for taking the time to help!

    • Ah hah.

      You’re right, it needs to be {{text:Kanji}}. Using triple braces is another way to prevent formatting and they have weird interactions with how other things in the expression.

      Thank you for being such a good bug reporter.

    • I’m makin lots of test decks and syncing them to AnkiMobile and sometimes they work and sometimes they don’t.

      So far, they’ve always worked if I’ve gotten them set up so that they are working on the desktop, checked the media database, then done at least one review, and then synced everything. And sometimes worked without doing all those things. Are you finding they’re not working even with doing all of that?

      • So I’ve tried a bunch of times and I still can’t get it to work. Are you syncing to an iPhone or a different type of phone?

        Any chance you could upload the png files? I posted this problem on the Anki Help boards and the response was that AnkiMobile might not support the svg file type?

        • This is working for me on an iPhone 4 running iOS 5.1, though it seems so quirky about what it takes to get it to start working that I’m not entirely sure what the necessary sequence of steps is. I’ve been syncing via Dropbox and AnkiWeb; you could also try the via iTunes method.

          I just remembered I’m running a beta version of AnkiMobile… I think the beta was just supposed to be bugfixes rather than adding new features though.

          I will upload PNGs, though it might not be until next weekend.

  7. I think what could be even more helpful are diagrams which color-code (and possibly identify) at-a-glance the radicals which make up a character. Anyone wanna take a crack at it?

  8. I got this to work (mostly) with Anki syncing via dropbox, but I have a strange problem with it. On my mac the contrast stroke order image shows up only as the numbers until I try and click on the image, then the entire image is visible as it should be. I have no idea why it does that. I haven’t tried it out via mobile yet to see if I have the same issue. Has anyone else experience this? Is it fixable or should I just start playing connect the kanji dots? That’s one way to spice up RTK…

    • That seems to have something to do with how these SVGs interact with Anki’s SVG display. I have experienced it when the whole diagram isn’t scrolled into view, and with some templates regardless of that. With those I have found that adding ” ” before the img tag fixes it.

      If it’s still an issue, when I upload the PNGs they won’t have this problem.

  9. Using GoogleTTS to generate readings! Well, I’ve got both kana and kanji – which one is better to use as a base to generate off of? It’s possible it could mispronounce the kanji, but it’s also possible it would sound less natural using the kana (bad word boundary categorization) and mispronounce certain particles. I’m not sure what to use…

    • As I write about in part 2, I recommend generating with the kanji and then fixing any that it gets wrong. I generally mass generate them and then only check them by reviewing the card; I mark any that are wrong and fix marked things later.

    • Despite I haven’t downloaded or googled anything yet, I’m pretty into watching some dubbed C.S.I’s new season episodes. I’m not sure whether if the show is too difficult for me, or if I’ll be able to get the subs but, I’m so excited about the possibility of inputing video on Anki, that’s gonna be good for everyone’s progress in my opinion.
      The funny thing is that, when Adshap first blogged about reviewing with images and sounds, I was like “that’s so cool”… I’m glad Anki’s been improving day by day and giving us more options to make our studies funner ;)

  10. For some reason i can’t get this to work with my deck. I have added all the sentences from genki textbooks and followed this guide, but the audio file just doesn’t have any sound to it. Does anyone have this same problem?

  11. I’ve been wanting to put audio in my decks for a while now but didn’t because I thought it would really slow down how long it took to make each card, but I finally got myself to give it a try. It’s really surprisingly easy! Even using audacity to snag lines of a song for my lyrics deck only took a little bit longer than making the cards, and the googleTTS only takes a couple of seconds. Thank you very much for the helpful guide!

  12. I have been slowly making my first 1000 sentence deck from the genki books and I have been getting the audio off of the cd thats included with the book. However I have just relaised I have been putting the audio on the question section. Should I change this? I thought it would be helpful hearing the sentence at the same time im reading it and then trying to remember the meaning of the word I dont know yet. Or should the audio be in the answer section? (Note: Im refering to 1000 sentences here only, I understand that kanji audio would need to be in the answer section.)

    Sorry for so many questions.

    • Should be in the answer section. Otherwise you can’t test whether you can read the sentence or not (since you will end up hearing it before you get a chance to read it out loud).

  13. Can this be turned into animated gifs? I think that animated kanjis or frames like in could be better.

    • This is just a thought I had, but I think gifs might be ultimately harmful, at least past a certain point. The reason is that as you get better you start going through your cards reasonably fast, and when you do have doubts about stroke order they’re usual very specific and clear points (usually about something like two strokes that cross each other but for which is not clear which is drawn first, or which of two parts of the kanji are drawn first), and having to wait for a gif file to run when you just want to check a single specific point might just end up being aggravating. (and, on another note, it would also be harder to harmonize with ankimobile, if you use it)

    • I don’t mean to argue — I’m sure it could be better for many people like that; personally, I like it how it is. It just takes one quick glance and I can tell exactly how it is written rather than following along or waiting for an animation.

    • Personally I prefer non-animated for the same reasons that other people posted. But I understand that different people learn differently and I can see that for some people the initial learning process would be enhanced by watching an animation.

      Translating SVGs into other image types in a cross-platform way on python is a bit tricky, but it would also be possible to animate the SVGs using Anki’s JavaScript support – and something interactive like that could also have a way to skip to the end when you’re past that initial stage. Unfortunately I’m really busy at the moment and it’s not a priority for me, but I’ll consider it.

      You might consider one of the methods of creating automatic links to web pages and visiting the page for your kanji and watching their animation any time you fail a card and feel like you’d like to see it animated.

  14. This has made a great addition to my RTK deck. Thanks! However, I am using the sixth edition of RTK, and so I have recently stumbled into a few problems. The kanji 嗅 (keyword ‘sniff’, Heisig no. 129/3008) is missing, and I suspect so are the other 175 new additions to the book. I would like to add these missing kanji to my deck, but I was unable to get your script to work, possibly because I am using Python 2.6.1? Anyway, could you please suggest a possible solution to my problem?

    • I just double checked, and that is included, so there’s some issue with your deck. (Could you have a space as well as a character on the card? Did you delete unused media before adding that card?)

      The addon for anki 2.0 should work a lot more smoothly, so I’m looking forward to 2.0 going stable.

      • I do apologise. The kanji is there, my Anki implementation must have deleted the unused media for some reason. By the by, I have noticed an error in the sixth edition relating to that kanji. (Just from my close inspection of it this afternoon). It appears that in the book, the kanji is missing the final stroke (the one the turns the 大 into 犬). Also, what do you know about Anki 2.0?

  15. I don’t want to sound like a “newb”, but could someone please tell me where the screenshot at the bottom of the article is from? I really hate it when people don’t put any sort of indication of where their screenshots came from (especially the cooler ones).

    On another note, I had thought that something like this would be a great idea. You know, translating episodes to learn the language better. It’s great that this technique was devised and I look forward to using it. The best learning is the type where we are tricked into it! :)

    • That’s Clannad, an anime that I think proves that you don’t need great subject matter to be a great show. Unfortunately I found after I got farther into the episode that the the timing of the Japanese subtitles got farther and farther from the video.

  16. I don’t know how to edit here. I checked that one out by it doesn’t let you choose tts engine, it does use like other 3 besides google’s. Pity cause misaki is simply he best jap tts engine. I guess I’ll have to stick to task automation to give voice to those sentences …

  17. I have downloaded the googletts plugin on my anki but I have a problem.
    I use the sort-keys to read the question field when I can’t remember how to read the kanji or after seeing the anwser to train listening. But when I do it, if the sentence is to long (more then 9-10 characters), it only reads the first part and stops. Even if I click again it only reads until a certain point and when i close anki it reads what missed (sometimes 3 or 4 sentences at the same time).
    Please help me!

  18. A great place to get audio + sentences is Japanesepod101. Though audio transcription of sentences requires a premium account, when you first sign up you get the premium account for a month (for 1$ and non-recurring).

    They’ve tons of material spoken by various native speakers, and every sentence has the audio transcribed in a manner that’s easy to copy n paste straight to Anki. It, at the very least, seems useful during the beginning phases when your choice of sentences is more dependent on getting used to the grammar etc.


  19. Thank you very much! After a lot of messing around, I finally got it to work on ankimobile 1.8.
    For it to work on ankimobile (not sure if this has changed in the new ankimobile as I haven’t yet upgraded) double quote marks are required:
    <img src=”{{text:Kanji}}.svg” />
    (Note the added double quotes around the filename.)
    Without these the images just won’t show on ankimobile.

    • It doesn’t always get kana words right, but it’s much better than just reading without any indication of pitch accent. It is too bad that there’s no way to correct it when you know it’s wrong though.

  20. I have a question. My plan is to put the audio in the question field.
    What I’ll do is that I will listen to the audio first and write the answer on a separate paper. Then after that I’ll check if what I write is correct in the answer field. Is this ok?

  21. Hm, I’m having pretty shoddy luck trying to extract these files on Windows. It looks like the files were given Unicode names, but any program I use on Windows – the shell default, 7-zip, or WinZip – all interpret them as ASCII and refuse to extract some of the files because they appear to have characters that are invalid in windows filepaths (like ‘:’).

    Luckily, I do have W7 Enterprise, so I’m currently installing the Japanese language pack to see if that helps…

  22. Just changed from the somewhat troublesome implementation of stroke order diagrams in Anki 1 to the Anki 2 addon implementation, and it totally blows the previous version out of the water (including on how sweet looking the diagrams end up being).
    The one critique I have is that you really should publicize it more on this site (it totally should be mentioned somewhere one the post above, if not even get it’s own post). I was pretty much lucky to find it before I tried to painstakingly address the problem I had one kanji at a time…

    • Thanks for the info. I added an update to the top of the post. If Cayenne or any anyone on this site is interested in writing about the new version of this for Anki 2, I’d be happy to put it up.

  23. Hey guys, I’m just about to embark on my Japanese journey so I downloaded Anki and got ready to start. However, once I discovered that I could add a stroke order add-on such as this, I immediately tried to figure out how. I have downloaded the Kanji Colorer into Anki, but after this I have no idea what to do. I am not tech savvy at all and I am extremely confused by the directions on the shared plug in at ( I also have a Mac if that helps. I guess what confuses me is in the directions when it says “Add a “Diagram” field to a model with “Japanese” in the name and a field named “Kanji”.” I don’t know how to go about doing this at all.

    Is there anyway someone could please give me an extremely basic description of what to do, or suggest something else that I can do. Sorry about the hassle and thank-you in advance.

    • I had trouble with this too.
      You need to rename the note type (“tools” > “note types”) of the card and add “Japanese” (ex: from “Heisig” to “Heisig Japanese”). Then add to that card type a field labeld “Diagram” (and don’t forget to add it to the back of the card, too).
      Then (re)genrate all the cards from the “tools” menu, and it should work.

  24. I find anime I want to use but the scripts never match up properly. Sometimes anki/addons/plugins etc can be very frustrating…

    • You can use Aegisub to shift the script file.
      Sometimes I’m forced to shift it differently for 2 sections of the file (or sometimes 3 or 4 sections for very long stuff), but I’m yet to find anything that is truly unusable. In fact, around 3/4’s of my daily new sentences currently come from using Subs2srs.

  25. Hey guys, not sure if anyone else is having this problem but
    I can’t seem to get subs2srs to work with Anki 2.

    I use 2 context lines, I’ve added the necessary fields in the correct order in my card template and everything is lined up on the import screen, but it still does not import correctly. (Yes I’ve moved the media files into Anki’s media folder also + enabled HTML when importing).

    Everything looks fine when I am in the browse menu, but when reviewing it just shows the tag ID of the card.
    Using latest version of subs2srs.

    • Well… by “Everything looks fine when I am in the browse menu” do you mean that the process is creating cards with the right fields all filled up correctly?
      ’cause if that’s it, all I can think of is that you created a card type with all the right fields, but forgot to configure which fields (and where) actually appear in the cards… To edit this go to “Tools>Manage Note Types…”, then choose the card type you are importing to, and select “Cards…”.

      • Sorry, I’m not exactly sure what you mean.
        But here’s how my card browser looks:

        The tag field is red because Anki 2 now needs the first field to be unique, but subs2srs makes it the same for each card. That shouldn’t make the card not work, though, right?

        For added clarity, here’s how one of the cards look when reviewing:

        • Ok, looking at your screenshots I’m pretty confident that your problem is what I thought it was. The problem really has nothing to do with subs2srs, all that’s happening is that you didn’t configure you card type fully in Anki (card types are what this article calls templates. I believe this terminology was changed between Anki and Anki 2).

          Ok, so the first thing you need to do is figure out what is the card type you are using in your “test” deck (I can’t really see this from your screenshots). On Anki’s main screen do “Tools > Manage Note Types…”. Just go through each card type you have and press “Fields…” until you find the one which has “Tag”, “Sequence Marker”, “Audio”, etc, as its fields.
          (by the way, you were probably here before when you created those fields in the first place)

          Anyway, once you locate the right card type, click “Cards…”, which will show you the problem to be solved. I expect that in the window your “Front template” box will say “{{Tag}}” and your “Back template” box will say “{{Sequence Marker}}”. Or in other words, right now Anki thinks that all you want in front of the card is the Tag, and all you want in the back is the Sequence Marker, which is of course silly of it.
          So all you have to do now is change those two boxes. On the “Front template” you probably want it to say something like:
          I’ll leave you to figure the “Back Template” (as I don’t know what you want to put there anyway).

          • Hey awesome, guess it did just need a bit of formatting! Thanks a lot!

            Also, I’m trying to get my cards to look like the one in the article above here. I can’t seem to get the right font though, does anyone know the font name of what’s being used up above?

  26. I have one question regarding this. How do you get the furigana on top of the expression? I can do this just fine when I create my own cards by clicking add using the Japanese type card. This is how I add all of my sentences from other sources. However, I can’t get it to happen when I use the card type that I created for importing (which I called subs2srs and copied it from the Japanese note type and added more fields). Is there a way to get the furigana to automatically add?

    • Okay, sorry for this ignorant comment. It appears there is a reason you put “Japanese” in front of the subs2srs note type! =D

      I can now get furigana in the card just like I do from other sources. However, it still doesn’t add automatically on import. I’m guessing this is just something I’ll have to deal with? It’s not too big of a deal, as I can just tab from the “Expression” field to add the line with furigana whenever I’m adding the meanings/definitions. But if there’s a way to have it be done automatically on import, I’d definitely like to know. Thanks!!

  27. Question, where can I get the deck seen in the first picture? It’s beautiful and it’s got the perfect background that I wouldn’t strain my eyes. I would pay for that! :P

  28. How is AwesomeTTS better than GoogleTTS? I looked at it, and it didn’t seem any different. The only TTS I could use for Japanese was GoogleTTS.

    • Let’s see…

      I think the most common ways to get the problem you guys are describing are the following:
      – you either forgot to toggle the “import HTML” box when importing the info into Anki or
      – you transferred the media into the Anki media folder after importing rather than before.

      Importing again (with the “import HTML” box toggled) should fix your cards, if I remember correctly.

      • Ah! You’re right. I read elsewhere about enabling HTML but I couldn’t find the box where it was, obviously because I’d already imported it. Thank you so much!

  29. Alright, I have almost everything perfect, just the way I want it…the only thing that confuses me is how to get the furigana for the kanji in the definition on the back of the card. I can get the furigana to show up for the main expression, but can’t figure out how I would get it to work for the dictionary definitions beneath it (よく調べて in the example in the article, for example.)

    Anyone have any ideas? Do I need to make a new Reading2 field or some such?

  30. Hello there once more!

    Well, I am feel a bit exhausted and desperate. I ordered a few animes which have japanese subtitles (Baccano!, Claymore and Log Horizon), with the aim of fusing both subtitle files with the respective sentence and a screenshot into an anki deck.

    So far, I just can’t shake the feel that I won’t be able to do it:
    Most of the subtitle files are asynchron from the ones from kitsunekko and have vaguely different timings. Seeing how my japanese is still pretty bad, I honestly can’t figure mid episode, which japanese phrase belongs to which english sub. And, with those few I do know, I can’t shake the feeling, that the english subtitles are mostly oversimplifying.
    I already invested now around 25 hours with basically no showable result.

    Has anyone of you guys a neat, simple anki-deck lying around they’d be happy to share? At this point, I don’t even care, what Anime it is, as long as I could finally understand some of it.

    ありがとうございます in advance

    • go to ankishareddecks. There are heaps of pre-made subs2srs decks.
      There’s even a persona 4 deck from memory! Have fun!

      The kitsunekko subs can be really irritating. Sometimes they are only 5-10 seconds off, sometimes minutes. For the new dbz movie they were actually in sync! Bit of a lucky draw with those bad boys.

      • You can use programs like Aegisub or Jubler to make the subtitles line up with the audio. It’s really easy to figure out, I’d recommend it because now I can watch pretty much any anime I want with Japanese subs and it has helped me immensely.

        Also, the reason they are 10 seconds off is because in the subs, they usually don’t include the 「この番組はご覧のスポンサーの提供でお送りします」transition between the theme and the show. If it’s a couple minutes off then its because the person who made the subs watched it on Japanese TV with commercials involved.

        • Thanks to both of you!

          Well, the Timings are not the only hardships: Some of the sentences are translated a bit… strange, sometimes, two People talk at once and the way the subbers put the subtitles differ… Trust me, I already tried fiddling with subtitleedit.

          However, a big thanks to the reminder for doublechecking with the anki shared decks. This could be awesome!

  31. Hello, I’ve got a question. How do you export based on tags? The only export option I can see is limited to the deck.

Leave a Reply

Your email address will not be published. Required fields are marked *