Mate: Netflix browser extension for learning words via subtitles
matetranslate.com
matetranslate.com
Edit: maybe not so niche! Someone build a WebExtension!
If I could, I would have the entire script as well, like Script to Screen does: https://www.facebook.com/pg/scripttoscreen.film/videos
One explanation might be if you're watching a 5.1 surround sound movie and playing it through your monitor's built-in speakers which can't handle the extra channels.
I work they use fewer words....it's unfortunate when it spoils a delivery several seconds before it happens.
If it's a comedy, there's not going to be any sort of quiet sounds, so I keep the subs off. If it's an action that usually has gunshots way louder than dialogue, I'm going to adjust the volume to those gunshots and turn the subs on.
The closed captioning settings on a Roku player hint at something like that, but I'm not sure I've actually seen it work. Instead of just off and on, they have 4 settings: Off, On always, On replay, and On mute. (More details: https://support.roku.com/article/208756848-how-to-enable-clo...)
Another possible variation on this would be to just let you scroll back through subtitles without stopping the video. It would be nice to continue without breaking the flow, but maybe it would be confusing or weird for them to be out of sync.
# ~/.config/mpv/input.conf (lines 20-22)
LEFT sub_seek -1
RIGHT sub_seek 1Edit: sub_step is also great to adjust subtitle delay instead.
Was nice when you couldn't understand the subtitle (slang that Google Translate doesn't know, for example). You'd switch to English for a second, go "aha", switch back, and resume video.
1-2 years ago Netflix introduced a regression where that feature doesn't work anymore. Now you have to go back and re-trigger the subtitles for the current frame for them to appear.
Sucks for language learners.
https://chrome.google.com/webstore/detail/language-learning-...
Will have it done by end of week :)
I used it to start trying to hear and parse Japanese (only about a month in of learning, so brand new, and just hearing the words and sounds of the language is still tricky).
However, creating a flash card with a translation to your native language is not the best way to master vocabulary. A translation (or, "gloss") can be helpful initially, but it's more important to see the word in context.
It would be great if a browser extension would include spaced repetition / flash card creation of both production cards (sentence with a blank usually) and comprehension cards (the word alone) and pull up the original video frame for context whenever checking one's recall.
FD: I hold an MA in TESOL / applied linguistics / language acquisition, and I have several years' experience teaching ESL students. However, I am not yet fluent in my own second language (Russian). Working on that with the Fluent Forever methodology (though not their new app).
If they have patience, it's good for anyone. My children have been able to learn many languages this way from scratch to a very high level of fluency that I only minimally speak. I look for children's tv network apps for their iPad from countries with the target language.
I think watching things in their original language is key. You pick up on things using context, some of which is lost in translation.
As an adult, I have way less patience to stare at a screen and try picking up stuff that I don't understand. I've tried finding some Dutch shows with English subtitles, but it was shocking how little content I could find.
There is an extension to Anki that does what you suggest with the video frames (at least for Japanese?), and I want to integrate that (or similar) at some stage. But there are heaps of cool ideas to implement if you take the flashcard database as a first approximation of a representation of the learner's (lexical) knowledge of a language and then enrich real-world texts that the learner is interested in with enough help to appreciate the real-world content. The beauty with having a representation of the learner's knowledge is that you can take a number of different approaches (say depending on the % of unknown words) or even just let the learner decide how much help they want/need on a given content consumption session. With even a small number of users the data generated will also be pretty awesome for research purposes.
It's pretty clunky still (I haven't started the thesis yet) but it is all going to be completely open-source and I definitely hope I'm not the only one to contribute to it, so hopefully it will get some momentum at some stage, and there will be heaps of clients that can talk to the server part. https://transcrob.es
I would watch the show with subtitles; when I didn't understand something, I would pause, look up the phrase/words, write them down for future repetition and continue.
A technology solution for this is awesome.
If you're learning how to read/write the language subtitles in that language seems like a great option. If you're starting out, it helps a lot to hear the language.
I can press a button and have the current subtitle analyzed for words with mecab, or press another button and have audio + screenshot + text sent to Anki for review.
mpv allows to seek previous/next subtitle, so even relistening a particular phrase is really convenient.
It makes language learning really fun.
The code is here https://github.com/pigoz/mpv-nihongo, keep in mind that a lot of stuff is hardcoded, poorly documented, and it requires an mpv patch to work.
If there's enough interest, I can try and make it simple for someone that's not me to use.
Also they have a subtitling community - any user can contribute to shows translation.
I am struggling to understand why Netflix is not thinking about that. It would greatly improve the experience and, probably, user base too.
It's activated with a keyboard shortcut: Cmd-E or Ctrl-E.
But these edge use cases will actually become less and less common as big copyright tightens its grip on how content should be consumed.
Like, how can you get this functionality on an iPad?
This is such a great idea. In my experience, most people complain the subtitles are distracting because they can’t help but focus on them and read them.
This is exactly what learner readers should be doing in front of a screen.
This is generally a good way to immerse yourself in a language with content that actually interests you - now you don't have to maintain a separate window/notes/etc.
I really like the idea, and it looks well executed.
Request: please add an option to not continue playing the video after I click "save". It is aggravating when there are several words that need to be saved at once and I have to pause the video quickly to get all of them.
You could also make the UI much smoother by copying the way the FluentU player works: pause the video and show a translation as soon as the user mouses over any subtitles. Unpause the video when they mouse away from the subtitles.
It would also be really helpful to have some contextual information saved with each word. The surrounding phrase would be helpful, for instance. It would be really fantastic to be able to save the screenshot (in low resolution to save space), but I understand if copyright prevents that somehow.
I really, really wish there were a more open solution for this. Has anyone else used LWT (Learning with Texts)? You could specify the dictionary API/URLs to use. (There were also really primitive word splitting regex settings, but I didn't find them very useful.) I want to be able to get more information on complicated words, from either compounding (German) or extensive morphology (Turkish). I want to be able to look up Chinese, Japanese, Vietnamese or Thai words; these languages require extra processing to extract words because they don't separate words explicitly. When studying Japanese or Chinese, I want to be able to save characters for later study, not just words.
In general I really wish I could dig in and add functionality that I know I would love but I'm not confident that a large enough audience to warrant core app changes would need.
Why they base it on your display language and don't just have a setting that is "show me X, Y, and Z language subtitles if available", I don't know.
I learned a lot of my early English from reading tabletop RPGs, so I've got no problem learning about twelve kinds of polearms before I learn the words you need to get by in a hotel…
(Totally getting this though---I might not be sold on the Netflix stuff but in general this plugin seems invaluable to me at this point in my Russian skills).
Disclosure: I'm one of the app developers. Let me know what you think :)
[1] https://deepmind.com/blog/wavenet-generative-model-raw-audio...
Similarly, how neat would it be to have a program that does this for object in video, and testing the viewer on the name of the object? Perhaps to avoid error as a result of imperfections in the neural net, it would require 2 parties, the learners and the teachers. The teachers would go through a video ahead of time with the video, making sure there are no errors, then that session would be saved and the teacher could play it back for the learners.
Copyfish works well, but since it is a generic solution for copying text from embedded subtitles/images, it is not optimized for language learning.
At $10 lifetime user price, it's not bad. But if you watch a couple shows to try it out and then want to save your work, you will have to pay.