Hacker News new | past | comments | ask | show | jobs | submit login
Language Learning with Netflix extension (soranews24.com)
458 points by davidzweig on Jan 13, 2020 | hide | past | favorite | 140 comments

Hi. Creator here. If someone from Netflix is reading this, we'd like to be in contact (email in profile).

Hell, here's my 15 minutes. I'd be interested in a job with electronics/software/mech engineering (ideally all three) in the US/EU/Australia (I can work in these countries). Some info about me on this post from earlier in the year: https://news.ycombinator.com/item?id=18809704

The other LLN creator here. I am looking for someone to correspond with or hang out with (in Europe), to conduct amateur inquiries into alternative approaches to AI algorithms (and perhaps discussing unusual interests such as lifespan extension and futurism). ognjen.apic@gmail.com

Your extension is great. Thank you! I've used it for a while. It makes something that I used to do, i.e. watching foreign language content on Netflix to advance a language, much easier. I only wish something this amazing were available for more services (or maybe Netflix needs to just license every show and movie ever made?!?!?).

Great work on this extension. Thank you both!

Hi there. Would you mind forking this extension to Firefox as well, please?

Related tip if you won't/can't use this extension: Netflix shows normally have many more subtitles available than you get shown (and sometimes audio tracks). The selection you see depends on your interface language and locale.

For example, I can't use Spanish subtitles for one show (https://i.imgur.com/BBWuuPs.png) until I change my interface language to Spanish (https://i.imgur.com/bsyYJGR.png).

I still can't get my head around how user hostile they made this. I mean, why? I understand showing the most likely languages on top, but why hide content that's available?

I have bilingual kids (Dutch and Danish, living in NL) and the amount of setup we have to do each week to get and keep Danish audio for cartoons on Netflix is pretty nuts.

Strangely enough, many countries also have cultural laws around what languages can be used for subtitles and audio tracks. For example, IIRC there is a country in SA that doesn't allow non-Spanish subtitles if the primary audio is in Spanish.

There are all kinds of permutations like that which can look like arbitrarily bad UI decisions if you don't know the greater context.

They're still pretty poor UI decisions if you're implementing misfeatures for every country based on the limitations of some countries. The more harmful it is, and the more money you have, the less justifiable it is.

I'm fairly positive, at the scale of Netflix, and the relatively minor amount of UI work they need to do, they aren't struggling to implement country-specific features. More likely than not, the ball either got dropped, or the misfeature wasn't understood as one.

I don’t think it is malicious - many people are overwhelmed when presented with more then a few choices, they are catering to that crowd.

I swear there's a subset of designers who are just blindly wrecking tons of what we collectively as an industry know about UX. "Overwhelming the user with too many options," or "Can't find the option I want" are both discoverability issues that design needs to take accountability for. Hamstringing your product by removing options because you don't know how to properly set up your discoverability is just insane to me.

If (say) I specifically want Danish subtitles, Netflix already knows that I use them, and I know they exist, that option is available to me nowhere.

I could see this as a reason to keep "you might also like ____" lists short, but for languages/subtitles I'm going to seek out one of the four languages that's even slightly useful for me to switch to.

Just keep preferences and not stop giving choice.

Keep thing simple but not too simple.

I’ve traveled around the world and have seen how insane the differences are with Netflix experiences. It’s interesting though, you can almost see how they struggle with licensing issues mainly, while from a consumer’s perspective it seems like issues with their competency or offering. You get lucky sometimes, like catching Better Call Saul 2 weeks earlier in Malaysia randomly, than the US release.

Yeah but it's Netflix, the company that plays audio previews at you with no way to disable the misfeature, so they could just be fuckups

I just mute until I've found what I want to watch :(

I bothered to write an email to them asking about other subtitles, than the handful I'm not interested in. Never got a reply. Would be really nice to watch some series/movies with, in our case at home, Russian subtitles. Or heck vice versa, Russian audio track with English subtitles, I honestly don't care anymore.

And I'm sure they exist, I remember their subtitle 'program'.


I really thought that would provide users with more subtitle options to choose from.

Not sure, but maybe the subtitle tracks have different licensing?

No, it is simply to avoid clutter in UI:


> There are too many languages to list them all in the menu and have the one you want be easily selected - especially when you are viewing on a TV with only a left-right-up-down remote. Accordingly, we show the commonly used languages in any country.

Well that's the first time I've legitimately gotten upset at a UX decision in a long, long time. My wife is homesick, 5000 miles from her country and can't watch anything in her language... Because they couldn't be bothered with an advanced settings UI anywhere? Seriously, Netflix?

Maybe that's an entitled attitude, but to me it's in the same ballpark as accessibility. It might not make a difference to 99% of users, but to the 1% of users in the long tail, it's royally screwing them for lack of a relatively small amount of effort. (In this scenario, a tiny amount of effort).

As someone said above, she should try to change the language of the interface to her language.

In my experience, that’s not a solution. I always use Netflix in English, but frequently encounter US content with only regional subtitles and no English track.

So set the UI language to watch she wants it to be? This is what profiles are for.

Prime video shows all subtitles. Even if we assume it is licensing, then Netflix's own shows should show more subtitles, but they just don't. I think they see showing tons of subtitle languages being shown as an UX issue.

So form over function then?

That’s also a form of UX fail.

It's available, you just need to jump through undocumented hoops to get it. I know not to underrate the media industry on this stuff, but I doubt they have licensing that says "make available but only through shitty UX".

But "hide from the UI" might be sufficient to pass legal review for a "make only available to users of language x, y and z" clause.

I could easily imagine this happening if they have two separate layers of availability checks. One general authentication check (likely with coarse "DVD region" baked in) owned by one suborg and another to handle all the eccentric subtleties of hundreds of different licensors owned by a different suborg. As laughable as "security by hiding the UI" is, when it's only protecting licensing nuances that have traditionally only been enforced by broadcast range (which is roughly continental for satellite) pressure to fix it can be very, very low.

Licensing might have given the impetus for the gating. And then the implementation is made like this, either as an oversight, or to provide plausible deniability to Netflix.

Here in Germany violence in video games used to be heavily censored. Much less so now, but when I was growing up, you basically couldn't have blood in games. I remember playing a GTA game that had bleeding NPCs or not, depending on whether you selected German or English as the game language.

> you just need to jump through undocumented hoops to get it

The hoops are documented: https://help.netflix.com/en/node/13245 ("Change your subtitle language" section)

It says "Due to the large number of languages available, Netflix only displays the 5-7 languages most relevant to you while streaming." and then proceeds to tell the steps to change profile language to get other languages.

Do elaborate on said hoops.

Wouldnt they need to check this based on location, and not on user interface language though?

Please don't do this. I can manage to read enough to do certain workflows, but I get so frustrated when Google presumptuously shows me dates according to the Buddhist calendar just because my IP is in Thailand. If my browser is reporting 'en-US' as the locale, the when available I expect to get that.

But you cannot expect to get unlicensed content, which is what your parent was talking about.

What makes you think everybody in a particular place speaks a particular language?

Shouldn't they just ask the user what their preferred languages are?

Because things such as Netflix are "boxed goods". You only get what's written on the box, and not a gram of extra value than that.

To better explain, with Netflix (any modern offering in fact) mentality one can't invent and market LEGO or even the PC or other general purpose toys. They want to capitalize on every bit of value from the product, and as a result, they can only give away "experiences" and not "platforms".

With a platform users generate extra value from which they can draw, with an experience it's just boxed, single-purpose consumer products.

I think having subtitles in your native language can hurt learning because it tends to be easier to just follow the subtitle than actually trying to understand the target language. You can't read both at the same time.

But I agree that Netflix is a great resource for learning a foreign language. I went from zero to being able to understand a normal TV show in the target language in one year, mostly by watching children's videos in YouTube and Netflix. In my opinion, what's more important is (1) the ability to order/group videos by simplicity after filtering it by target language (even children shows vary a lot in difficulty and for learning it's very important to choose the right difficulty for you) and probably (2) the ability to slow down the playback speed like YouTube.

I agree. That said, my wife subtitles every movie which normally I don't care about, except the subtitles are -wrong-. I don't know why they do that, it seems rather innocent, but drives me crazy as an English speaker. It's usually something not even so consequential..off the top of my head like replacing 'Time for dinner' with 'Time for supper'. Not an exact example, but it's usually along these lines.

It was said in a major Dutch newspaper last year that Netflix seriously underpays/over pressures the translators:


Basically in “the good old days” translating/subtitling was a serious job for serious money. But it has been turned into almost a “work from home, gig economy Mechanical Turk” thing.

And the quality obviously suffers.

I won't even try the shows whose trailers have the problem stated in this thread.

It's blatantly obvious 3 seconds in. Very unprofessional.

That drives me crazy too! Some shows are pretty faithful, but others are way off. The otherwise-wonderful Cuatro Estaciones en la Habana is especially bad -- it replaces all the local Cuban slang w/ standard Spanish, e.g., they'll subtitle imbécil when the dialog is comemierda. Why would you do that?!

Also strongly agree w/ GP; in my experience, you've got to hide the english to learn anything. Otherwise your brain isn't working in the target language.

Are the shows you're watching dubbed?

Was recently watching How to Sell Drugs Online (Fast), which is a German show originally, and the subtitles were off in a similar way. Drove me nuts!

I ended up switching the audio language to German and watching it with English subtitles, was a miles better experience for me.

Nope, original English shows. What's frustrating is that my wife uses subtitles for learning how words are spelled, but the words spoken aren't the same as the ones written!

From what I've heard, when dubbing shows they normally adapt the translations for the dub to fit better with the character's mouth movements and to make it more 'idiomatic' in the target language, but the subtitles keep the literal translation. It would be interesting to learn why, I find it drives me crazy as well.

Semi-synced dubs are an early example of uncanny valley

This has been an issue with Netflix for years. It's almost as if the translators of movie or TV series don't have access to the videos themselves because every other sentence is slightly wrong.

Just tried it out with a German show I watch. I think the issue is that the native language subtitles (English for me) can have subtly different literal translations than what might be on the screen in the original language. So while the English subtitles can convey the same meaning, the word-for-word translation is going to be off. This is fine, really, and any language learner is probably already aware of this.

Otherwise, the implementation looks great. Easy, simple to use and intuitive controls. The running list of subtitles on the right is nice. To your point about playback speed, maybe it's just me but I like the challenge of trying to keep up with the foreign language and it helps with developing my listening skills (I also really don't want to watch children's shows). I think the ability to toggle the Auto-pause more or less solves this problem, at least for me. Would still like a Firefox extension though. This will be the only instance in which I use Chrome.

Edit: Hmm reading the other comments here, maybe this is actually a problem with the translation itself? If non-native speakers are noticing instances where it's off it seems there might be a quality control issue.

Been using this extension for a bit now. Turn on machine translations. The default translations are translating the meaning but not word by word. The machine translations are the latter. Which is way better for learning, which is why I assume it's a premium feature.

I respectfully disagree, based on anecdotes. The Dutch-speaking part of Belgium subtitles everything, except for shows meant for kids <8 years old. The French-speaking part on the other hand dubs everything, even interviews in the news. Guess which part of country sucks at speaking English.

My point is you need to watch the target language audio with no subtitle in your language.

If the choice is English-audio + Dutch-subtitle vs French-audio only, then of course there is no advantage to the second group. Try restricting the French-speaking kids to watch shows with English audio only.

Which language did you learn?


Most effective Netflix kids/otherwise shows for this, in your opinion? TIA

The shows that interest you. The ones you would actually watch.

I imagine this rules out kids shows for most adults. It sure does for me.

Also start reading books asap. I spent way too long watching dub+sub'ed Netflix in my target language. Wish I'd tried reading young-adult books sooner. If you're going to be pausing on every subtitle frame to read it, you might as well try books.

>> I imagine this rules out kids shows for most adults.

Not necessarily. Maybe if you understand everything, they become boring, but to me watching an interesting enough show where you understand 50%-80% of the dialog and try to understand the rest by context is kind of fun.

It depends on your level. I'm a big fan of Peppa Pig, because I think it's funny/entertaining enough for adults who understand >50% of what's being said. But before you reach that level, you need to watch something simpler. I'd suggest "Superbeginner" and "Beginner" videos in a YouTube channel called Dreaming Spanish

Watch in a foreign language with foreign audio to increase your listening comprehension in the foreign language.

Only watch in your native language/subtitles if you want to comprehend the plot more.

Source: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4927148/

Well, the linked study suggests watching with foreign audio and foreign language subtitles, not merely with foreign audio alone.

My experience as a language learner (French) supports the study. When I use English (native language) subtitles, my listening doesn't improve at all. However, my improvement with or without French subtitles depends on my level of comprehension. If I can understand most of the show without subtitles (~60% or more) I improve faster without subtitles. If I understand half or less, I improve with the subtitles. I guess this is evidence for Stephen Krashen's theory of comprehensible input.

I've been wanting to learn Chinese, but I'm not sure Chinese subtitles would help that much compared to if I were learning a European language. Maybe if the subtitles were in pinyin, but I don't know if that would even be an option for a lot of media.

Watching Chinese TV shows with Chinese subtitles definitely helped me a lot, not only because the subtitles gave me a second chance to get what I missed (which is the same with other languages), but also because I could learn characters for those words I could already recognise in speech.

Watch in your native language only if you don't care about acting at all.

I can see several problems with this.

1. Japanese subtitles are completely different from what is said in Japanese dub. Subtitles are optimized for readability and use completely different phrasing, or even terms (for example dub says クイーン ("queen"), sub says 女王 ("jo-oh")).

2. Japanese subtitles and English subtitles are also different, because many lines have to be completely rephrased in order to express them in a (very linguistically distant) language.

3. The furigana (pronounciation hints) in these screenshots is not correct. "困った" is "komatta" not "koma ta". "妻は" is "tsuma wa" not "tsuma ha". If only they used ichi.moe's engine... (disclosure: I'm the author of ichi.moe).

The way the "koma" is placed above 困っ makes me suspect they first tokenize the text, get furigana for each segment (most tokenizers have this built-in) and then derive the romaji without correctly handling the sokuon っ when it appears just before a segmentation boundary. Usually it's represented in the romanization by doubling the following consonant, but of course that doesn't work if there's no following consonant because each token is treated as a separate string. They do get it right for "erikku" where there's no segmentation boundary.

If they're having troubles with correct romaji transliteration (and even if not, to be honest), I don't see why they don't just transliterate it in furigana then. The source data of Tokeniser dictionaries such as NAIST JDIC isn't even in romaji, so if they are using a proper tokeniser, then they're actually doing an extra step and throwing away data to transform it into romaji form.

The feature is probably targeted at absolute beginners who don't even know kana yet. The screenshot of the settings does show the option to select a different transliteration, but romaji seem to be the default.

That makes sense.

Hopefully it can be fixed, since geminates are one of the most difficult features of Japanese pronunciation for native English speakers to master. Erasing it from the orthography intended to aid beginners certainly doesn't help.

> (for example dub says クイーン ("queen"), sub says 女王 ("jo-oh"))

My Japanese comprehension is still very minimal, but wouldn't this be pretty normal? It seems common in manga at least to use kanji for semantic meaning and annotate it with furigana for a foreign word or in-universe term.

This is slightly offtopic but ichi.moe is a wonderful tool, thanks so much for making it.

Also check out the lingq exporter [1] which works with Youtube as well. Does something similar but also allows you to keep an inventory of your current vocabulary on the lingq platform. I'm a huge fan.


I looked into Lingq, watched a few short videos on it, and saw an interview with the founder, but I'm still not sure I get it. The site is mainly a text-based vocabulary acquisition tool, right?

Yes it is very far from intuitive (as a developer I can tell when uis are designed solely by developers).

The way I think of it is this: you do all of your foreign language reading through the app. Lingq keeps track of which words you have seen: when a new one comes up it's highlighted in blue, when you define the word (mostly automatic) it turns yellow. Once you mark a word as known it is no longer highlighted.

There are SRS features built in, but I ignore them and just add words to anki when I feel like I need to.

If you love reading it's a really great way to get language exposure. Since looking up word definitions only requires a tap or click it enables you to read above your level (like reading with training wheels on). And just seeing a word over and over again in different contexts is a much better way to learn than hammering away at flash cards. The import from netflix/youtube stuff as I mentioned is great as well.

I also upload all of my own material (mostly ebooks) and ignore a lot of stuff they have there.

I see this littered all over the comments:

Subtitles are not transcripts. They don't match word for word, mainly due to space and time constraints.

A lot of amateur subtitlers time the subtitle with the time of the spoken word, making them almost unreadable. Long text that pop up for a short time does not help convey meaning.

You want a short text, that anticipates what is being said. The reader of the subtitle is after all watching a movie, and might have their eyes elsewhere - They are not reading karaoke lyrics!

I agree with the author's assessment that having the kanji, the furigana, and the English is really helpful for learning.

But I couldn't help but notice that the the software didn't romanize 困った correctly. It was "koma ta" when it really should have been "komatta".

And I wish I didn't have to use Chrome :|

Request a Firefox version. It might not be hard to port.

Yeah, you can basically automate WebExtensions porting. There used to be a browser extension that did it for you.

I've wanted this literally for two decades. Thank you to the creators for finally making it.

> When you’re starting out learning a language, sometimes you can’t quite catch what people are saying. They’re too fast, or they use words/grammar you’re unfamiliar with. Being able to see the subtitles and hear the dialogue and see the translation all at once is a dream come true.

I’ve tried this but the subtitles often use different wording than what is spoken (when both are in same language) so it doesn’t work out well in practice. Maybe depends on the language.

An observation: in netflix, when you choose a foreign lang for the audio track as well as subtitles... they don't match.

By this I mean, let's saying you're watching Orange is the New black... and choose French audio and French subtitles. You would expect the subtitles to be correspond with what you're hearing.... they don't!! Kind of absurd and annoying. This is true across all shows it turns out.

You might find this helpful: http://languagelearningwithnetflix.com/catalogue.html

"Netflix usually has good subtitles for the original language of a title, that match the audio track. Subtitles often don't match the audio for dubbed audio.

The catalogue lists titles by their original language. It allows you to find French films, to study French, and German films to study German etc. By using one of the listed titles, you should have good subtitles that match the audio."

This can be an advantage when your level is just a bit higher.

Since you hear the words and see a similar way of saying it at the same time. This (in my experience) is usually enough context to parse the audio.

The net result is then that your brain gained 2 pieces of input at rate of 1.

You discover this early on when picking up a language. It's so pervasive and so annoying for language learners but I can understand why it happens.

If you're learning a language, CC are p much the way to go

My biggest problem with understanding french is being able to comprehend it at speed. I'm much better understanding song lyrics (because they're often sung much slower than speech.)

I just tried this with "A Very Secret Service" and It's immediately easier to understand what is being said. I can see this becoming a very useful tool for me.

As a strange coincidence youtube recomended me this [1] just after opening the article:

[1] Alexandra Stepien - Netflix and chilll like a boss: How to learn languages the lazy way - PG 2017 https://www.youtube.com/watch?v=-5d9JS_fE6E&t=1220s

I asked for this on HN a few weeks back. https://news.ycombinator.com/item?id=21853163

I really hope it supports Hanyu Pinyin. So excited.

It's not really what you asked for, but subtitles that you download from websites like zimuku.com or zimuku.cn have dual language versions: separate English and Chinese flag means the ZIP file contains separate language subs, pirate flag means it has one set of subtitles containing both languages - super useful if your video player doesn't support displaying more than one set or you don't want subs at the top of the screen).

Just tried the extension, and it seems Pinyin is not supported. There is no "Transliteration" option, like there is for Japanese.

I guess this is most useful if you don't live in the country you want to learn the language for. Native English speaker living in Germany here who speaks fluent German, but still lives most of his life in English. Movies are a good way to keep up language practice, but there's no way I'd want to dub "Marriage Story" or whatever in German. There are plenty of German films on Netflix which I can simply watch with CC German subs, if needed. At any rate, it's still a cool idea, particularly if you lived in, say, Spain, but wanted to see German subtitles.

Be warned that one of the worst ways to learn a language is through translations. Most translations leave out a bunch of information because between connotation, denotation, and grammatical correctness you can really choose one or two of the above. That's in addition to the fact that most translations are either produced by machines with no understanding of the content or even of the complete grammar of the target/source languages, or mechanical turk style low wage laborers with tight deadlines and questionable knowledge of either languagr.

Really cool tool. Honestly the catalogue sort alone on the site was useful for finding which shows on Netflix even have Korean subtitles.

Can anyone recommend any movies and/or series for a complete beginner learning Chinese from duolingo? I mean, ideally probably aimed at literally early primary school kids... some animated series, or Tom Sawyer -alike, or a drama about a lost puppy is what I brace for... even better if on youtube than Netflix, but the latter is also acceptable...

Try "Where are we going? Dad" on YouTube. Also HelloChinese is much better than Duolingo in my opinion.

What would be needed to make this off-line capable with supplied media? Just a bunch of well-formatted SRT subtitle files?

Obviously some kind of lookup between languages too, but what i'm asking is : is this just a fancy srt/language api parser?

I want to use this, but I don't use Netflix. It looks very cool.

Posted above as well but mpv user script that works offline: https://github.com/oltodosel/interSubs

Ognjen (guy I worked with to make this extension) previously made something similar for offline media (https://github.com/oaprograms/lingo-player). It's neglected now though and probably suffered bitrot.

We could use the code from our extension to support offline media.. but Chrome doesn't support MP4 playback IIRC. It's possible to make an Electron/NW.js app, but the vlc plugin isn't working with latest versions, last time we checked. Appreciate any info.

You could use the .ass format for fancy subtitles. Players like vlc, mpc, mpv support it. That would give you the subtitles in both languages plus the highlight of the words as they are spoken but clicking on the word to get a definition would need to be built into the player.

A: this is fantastic. B: why don't all streaming services simply have a by-language option? It's kind of a guaranteed "long tail" for any audio or audio-video presentation: people who want hear language X language spoken.

You can search for a language on Netflix though, e.g., search for "Italian" and it will show several tags next to "Explore titles related to", like "Italian-languae Movies & TV".

Somewhat offtopic; I wish extensions would work on mobile browsers... I watch Netflix on iOS or streaming from iOS to chromecast.

I see it is recently somewhat possible to run extensions on Android via a custom Chrome install. Does that work well?

You can also use Firefox for Android, which has extensions. Android Preview is much faster, but it doesn't have extension support yet.

> I watch Netflix on iOS or streaming from iOS to chromecast.

If you streamed to a Chromecast it wouldn't work, as all the cast protocol sends from the iOS device is the URL of the stream and the authentication key. The Chromecast fetches any subs etc itself.

I use all chrome extensions via Kiwi browser on android.

It's so sad that Google has deliberately disabled the support for it on their own browser (they don't want adblock maybe?) even though and independent dev has seamlessly enabled the same in his browser.

For this extension, Netflix doesn't allow Chrome mobile or iOS browsers (I think) to use a browser-based player, but redirect you to their app. Netflix are probably looking at user-agent.

This is nice, but all I really want is the ability to nicely see both English and German subtitles simultaneously. This is overkill for me and requires me to use chrome. Is anyone aware of a better solution for me?

I second the mpv-based solution, as suggested by st1ck. If you prefer vlc, it now (as of the beta 4.0.0)[0] also supports dual/secondary subtitles. Making it work seems a bit fiddly: first you need to turn them on (under Tools > Preferences > Subtitles/OSD > Dual Subtitles (at the very bottom) > Align — change to anything but unset, and possibly also adjust offset). Then, when playing a video, to select them for that video, you need to "Toggle secondary subtitle control" with Ctrl+Shift+V (this means that the normal subtitle control shortcuts like "v", "Alt+v" etc. now apply to the secondary subtitles) and press "v" the right number of times to switch to the subtitles that you want.

(Obviously vlc and mpv will only work for DRM-free videos, e.g. from youtube or a DVD.)

If you were asking specifically about Netflix subtitles, there used to be an open source NflxMultiSubs extension[1] for both Firefox and Chromium, but it was broken by Netflix introducing changes to its video player and discontinued. There is an active, open-source dual-captions[2] extension, but it's for Chrome only. (However, since it's open source adapting it it for Firefox should be straightforward.) Finally, as an alternative approach, you could try a Firefox addon which allows loading arbitrary subtitles in the SRT format to netflix,[3] and which might perhaps allow you to have both netflix's subtitles and your own SRT ones at the same time.

[0] https://github.com/videolan/vlc/blob/master/NEWS#L30

[1] https://github.com/dannvix/NflxMultiSubs/

[2] https://github.com/mikesteele/dual-captions

[3] https://addons.mozilla.org/en-US/firefox/addon/netflix-srt-s...

NflxMultiSubs is still working. I wrote a patch to fix it, and took over maintence on Chrome.


> NflxMultiSubs is still working. I wrote a patch to fix it, and took over maintence on Chrome.

That's great! It's a shame that it's (apparently) no longer open source, though obviously given the MIT license you're allowed to stop disclosing the source.

Thanks for the shoutout! :) - mike

    mpv --sid=auto --secondary-sid=auto --sub-pos=95  --slang de,en movie.mkv
Same works for youtube URLs. If it doesn't work, changing "de,en" to "en,de" may help.

mpv is so great. It was an immense downgrade to start watching movies on my ps4 with media player, netflix, plex etc. They(mostly media player) have issues on the very basics like pausing and playing video, subtitles, continue watching etc.

With a computer running mpv and plugged into the TV everything just works and I can trivially fix common issues like subtitle/audio sync with my phone as the remote.

On-topic: mpv seems to have a similar language learning extension to OP https://github.com/oltodosel/interSubs

I've made my own mpv extension for Japanese subtitle analysis too. It can also create Anki cards automatically (with screenshot and audio).

Having a scriptable media player is really something else.

This exists, albeit only for YouTube rather than Netflix: CaptionPop https://www.captionpop.com/

Creator here.. there's netflixmultisubs extension which is more minimal, which we also maintain now. There used to be releases for firefox, but we haven't been publishing them, just becuase we're not familiar with the process and lack time. LLN can be used in a minimal, non-instrusive way however, and has a pretty good algorithm for aligning the start and finish times of different subs tracks. netflixmultisubs shows the two tracks with their independant timings.

Ef yes! Been looking for this, can't wait to try it.

There's a similar product for YouTube videos called www.fluentu.com. I used it for French and thought it was great.

animelon.com was the first company I know of to do this sort of thing. They went through an accelerator where I was working working.

This is just amazing. Thank you!

Wow, this is very cool!

In practice, the foreign subtitles do not line up with the foreign words as well as the native subtitles line up with the native words. It took me years to figure out why this was.

In a conversation I had with someone who knew the business, the new language subtitles don't line up with the new language audio because the new language subtitles are translated from the original language subtitles, not the new language audio.

Why? This saves a step from someone having to rewatch the final new language dialogue and transcribe the new language as it is exactly spoken. That step can be skipped and the new language subtitles can still be translated from the original language subtitles, although they will be quite off much of the time.

There's practical reason too. Translations made for overdubbing are designed to roughly take the same amount of time to say, and ideally to carry the punch in the same part of the sentence (so it fits with when the actor makes the corresponding funny face). This is not always trivial, to the point that sometimes translators change the meaning entirely (especially with eg comedy - entire jokes are reinvented because they were untranslateable so they just put a different joke in).

Subtitles are translated with brevity and ease of reading in mind. Translators typically take more freedom with nuances and emotion because, well, you're hearing the actor say it out loud in the original language and that's where you get the emotion from. So it's ok if the subtitle has it in a different order / different "punch", etc.

I'm also pretty sure that translated subtitles have as their primary target audience people watching the movie with the original audio. Taking the two step process of going from the source audio, to translated audio, to subtitles seems like a needless detour for that (both organizationally, but also qualitatively).

I know this too well as I tried learning French by watching FR dubbing with FR subtitles on Netflix.

There's absolutely zero resemblance between the two. Different words used, different order within the sentence (dubbing says "A B C" while subtitles say "C B A" half of the time). Pure madness. I immediately had to stop because it's been impossible to follow the two :(

Practically, not many people want dubbing and subtitles at the same time, so it's probably rational choice to save the costs.

I also ran into this issue with French specifically, and got frustrated and gave up. Even original language French films/shows seem to have spotty subs (or none at all!) Most of them have English subs, but that's not as helpful for learning. Did you ever find alternatives?

> just put a different joke in

I wonder how often you get a different joke on the translated audio and yet another different joke in translated sub titles. Kind of wild.

"Maybe it's a drill." Said stormtrooper, when Luke and Leia hid and siren was on background.

Finnish translation was originally a physical object drill, and not drill as a practice .

This is like that time "Boring machines" are often mistranslated as uninteresting into many languages.

Yeah this is unfortunate. As a native English speaker living in Finland - going to the movies it's interesting seeing how dumbed down the subtitles are. Dialects, slang: a lot of it can't be translated directly. I just installed this extension and tested it on "Salt, Fat, Acid, Heat" and one of the first things I noticed - two lines into the show was:

"... that she grew up eating in Iran" => "joita han söi Iranissa"

The Finnish translation here is "that she ate in Iran". Contextually, it still makes sense, but it's important when learning a language why it's okay to translate it that way. Literally it means something different, contextually it means the same.

I might give this a try, with the auto-pause feature.

For a language learner, shows where they match up are gold!

If anyone is studying Spanish, the animated show Star Wars: The Clone Wars have audio that exactly matches the subtitles (Start Wars: La Guerra de los Clones). Good show with a good mix of action / dialogue that I learned of Spanish a lot while watching.

For Netflix shows, take a look here: http://languagelearningwithnetflix.com/catalogue.html

I would love a service that created matching subtitle tracks for dubbed audio for this exact reason

Thanks for this tip! Working on my Spanish as we speak!

This can reveal some interesting insights into how the film was made. I often notice extra lines that got edited out to make the scene more concise. Sometimes the director changes a whole shot.

That would be extremely frustrating. Due to certain sensory issues, when possible I often have the (English) subtitles displayed while watching a show, as I'll otherwise just miss the odd word or phrase.

If the subtitles do not exactly match the dialogue, it is very frustrating. This happened a lot with some older shows I have the DVDs of, where the subtitles are... abridged versions of what is spoken. Impossible to use really unless you are stone deaf.

But why are the original language subtitles not accurate ? and I am referring even to "simple" local TV series that didn't undergo several rounds of editing.

As part of my Swedish learning I watch a lot of Swedish shows with Swedish subtitles, there are always discrepancies

If there is a script for the movie subtitles will often be identical to the script even if the actors deviated from it. Much simpler and quicker to do than having somebody transcribe everything that was actually said and then having someone watch everything again to make sure it was transcribed correctly.

Isn't it because subtitles are typed in live? So they have to skip words in order to convey the meaning in time.

I imagine the audio is transcribed before voice actors speak; they aren't translating on the fly.. it seems to me the natural solution would be to have the transcription the voice actor uses be the same as what's provided to the subtitle generation.

This seems to require two translation attempts; the main benefit being they can be done independently/parallel, but at the cost of duplicating the work. But it doesn't seem to save any work (assuming VAs work off a written transcription/translation, presumably completed by other staff members)

FWIW subtitles for the original language will also sometimes vary from the original audio as well. I tend to re-watch shows on low volume with subtitles on so I've caught this a number of times, but it's fairly rare.

Usually I've seen the subtitles shorten the words in the audio. "You must come with me now" -> "Come now" or similar. I've been assuming it's to help the reader keep up with some quick exchanges.

> the new language subtitles don't line up with the new language audio because the new language subtitles are translated from the original language subtitles, not the new language audio.

Foreign DVDs include two subtitle tracks, one to translate the English dialogue and another to transcribe the dubbed dialog. At least, this is the case for my Chinese Disney DVDs.

My girlfriend has dyslexia and subtitles are a problem for her to read quickly. She would love an extension similar to this that “auto-dubs” by having a synthesized voice that reads the subtitle content and overlays this slightly louder than the original audio. Any ideas if something similar exists?

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact