Youtube Extension - Now Online!

Ognjen’s checking it out, yeah, you go to 40,000 items, more than we designed for. :grin: Will make a fix.

1 Like

Lol, that is nice to know the count, yeah, I have been a busy bee… :upside_down_face::honeybee:
Cool being the first.

I would suggest setting a limit, if any, of something like 30K per language studied.
It makes little sense setting a shared limit for all languages in the user account.
That would be a much harder to reach limit.
I appreciate the quick handling and your responsiveness as always.

In case a limit is reached again in the future, alerting the user about it would be a good precaution too.

Btw, you guys had a recent mention in a big Youtuber account of a German language teacher that I follow. She is a big deal.
In case you haven’t seen it yet, it’s kinda recent (from 2 weeks ago).
She has 611K subscribers so that would be great exposure for you guys.
You guys deserve it! Enjoy the fruits of your hard labour.

1 Like

There is a conflict in creating Youtube playlists while using the extension, because of keys reassignments by the extension, which prevents you from naming them properly.

I’m currently using Youtube playlists in order track my overall watch time, per language, so I’m adding every video that I watch to a playlist. I’ll suggest adding a similar overall ‘watch time’ tracking to the extension. Tracking all the videos you ever watched, of course, would also be an awesome feature (reviewing old content you studied).

Esperanto is lacking support. No dictionary, no color coding.

Here is an Esperanto Youtube playlist with over 200 subtitled videos:
https://www.youtube.com/playlist?list=PLoszEcyUpiSmwAMrbQ7Y-RosJVbnwBg4i

Many of his videos also feature dual subtitles support (check other playlists too in his channel - many subtitling options available). I have learned the language with him from scratch like that. The best resource for that language out there.

I’ll note that some languages, such as Swedish, Arabic, Esperanto and Greek, for example, don’t support speech recognition on Youtube, so you would be forced to deal only with human made subtitle tracks with those.

Often, in such cases, finding content that you can actually use, would be severely much less available. Anyway, you pretty much mostly solved that problem already, with your upgraded search functions.

There can be a big font size difference between different scripts with the default values.

For the latin script ‘normal’ font size is ok, but when I switch to Arabic, font size is too big and I need to reset it to ‘extra small’ to be an equivalent size.

Font size preference is currently a global setting, which means that I keep needing to switch it back and forth. It would be convenient to save a different font size preference, per different script, or per language, and to adjust the default values as well for each new script for optimal viewing.

There isn’t enough space for some words to display fully, in the sidebar, at the bottom of the line.

Btw, some languages i know well enough, that i would prefer to use the sidebar alone, without anything displaying on the video itself. I prefer it clean.

It would be nice to have that display setting to enable independently only the sidebar.

Thank you! Here’s the link to the Dualsub extension if anyone else needs this:

Their ‘Chinese Learning’ display mode is designed to move the dual subtitles up above open captions on screen, so stops the subtitles overlapping the LLY subs, which is nice! If you change LLY to show transliterations only, you can even have quad subtitles!

There’s still plenty of messing around with window sizes, but this is a really good stopgap until these features are added to LLY. Thanks again!

1 Like

Hey, sorry to bug you, but is there any estimation when that fix would be available?
In case it would take time, can you perhaps dump my English database, in the meantime, so I could continue study for a while until then? As long as it ain’t posing a risk to losing the rest of my data…
There are about 7,000 entries in English so that would give me some breathing room.

Suggestion: Sometimes the subtitle language is mislabeled on Youtube (or 2 languages might be in the same subtitle). It would be useful to allow the user to personally set the correct language of the transcript in such a case.

Just a heads up, you can change the primary subtitles on youtube the normal way, through the Youtube playback menu.

Thanks, will fix very shortly.

I know. I meant something different. Like when a subtitle track would be labeled as English but it’s actually Dutch, or something similar to that situation. User/Machine metadata errors (Youtube is not entirely as predictable as Netflix, since it’s community based).

Well… I can’t say such issues happen often, but it might be useful to have that manual control in case the issue arises. I think I encountered a few situations like that in the past where I could have used that manual control. Anyway, I’m not entirely sure that is really necessary… Just something I remembered that I thought about it once so I mentioned it.

The more common use case would be when a subtitle track has multiple languages within it, I guess, such as when a foreign language teacher, speaking in a foreign language, teaches another foreign language. That can happen. So if you have multiple foreign languages in the same track you can only mark 50% of it.

I’m a fan of laddering from 1 foreign language to another foreign language, which is why I tend to encounter such situations sometimes. If I could manually tell the plugin which language to consider the track as, I could then mark the subtitle 100% by parsing it as the 2nd language on the 2nd pass.

Polyglots videos might also have that issue sometimes (rarely).

Anyway, consider this a low priority suggestion.
I myself not entirely convinced that is so necessary, but it can’t hurt having that option…

P.S: Youtube’s speech recognition might also have erros in detecting the correct language (that is a much more frequent issue), but since the language was detected incorrectly the output is also not that great… (thinking Spanish is English, for example, and analyzing the Spanish audio as English). But since the generated output would be pretty bad, as mentioned before, I don’t think it matters much in this case. I just mention it here for completeness.

I increased the limit for the number of items. That should be fixed now. Please let us know if it works OK.

It sure does! Thanks a lot. I’m as happy as Larry. :blush:

Custom user’s dictionaries are a global setting at the moment, while they should be made a language specific setting. Not all dictionaries online are multilingual dictionaries.

Edit: A button for disabling the auto-scroll in the side panel, temporarily, can be useful. You sometime look ahead in the transcript and it pulls you out of position.

Edit: This sometimes happens (the white part). You can color it in a different color in this example.

Edit: Right mouse button currently toggles between yellow, green and white, which is quite useful, but I wish there would be a setting, which you could switch to, that would allow you to include the blue color as well. Like a 2 modes setting.

Edit: The white word is unmarkable due to the punctuation marks.

Edit: DeepL is a competitor of Google Translate, which some people say that it gives better results. Might be worth checking it out and adding support for it as an alternative option.

Notice here the word “coníferas” in two different colors.

From this video: https://www.youtube.com/watch?v=mNpJmoAWnFI

Another example from the same video with the word “porte”:

Again with the word “cedo” (same video)

“churrasqueira”

I checked the example with ‘porte’. One example is correctly recognised as a noun, lemma ‘porte’, the other incorrectly as a verb, lemma ‘portir’. I don’t really hold it against the linguistics code we use… it’s rather that we are feeding it sentence fragments, with no punctuation, so it struggles. Youtube doesn’t break up stuff into nice subtitles, it’s just a goop of words grouped by some crude algorithm. I wanted to have a go at recombining the goop into nice subtitles myself, would need some clever machine learning stuff, haven’t had time to work on it yet. However, youtube is now grouping translations into bigger subtitle blocks, so regrouping also means we need to create new machine translations. :confused:

I know. I meant something different. Like when a subtitle track would be labeled as English but it’s actually Dutch, or something similar to that situation.

Hmm, ok. Noted. :slight_smile:

DeepL is better in general, I think, for the languages it supports. But also pricey. We use Microsoft translate, which is decent and better value, and still about a third of income from subscriptions goes on machine translation API costs. We have something in the works though that may help.

P.S: Another idea, for starting to learn a new language from scratch using your extension, it would be pretty damn cool, if you could auto-mark (highlight) the first 1000 (up to 3000) most common words to the user, so he would know what to focus on and not get lost in the sea of it. I think you guys might have already implemented such or similar feature in one of your first releases, if i’m not mistaken.

I added the backend code for this feature over the last few days. We used to have this feature, it got taken out during a heavy refactor. So, the way it will work is you set your level in settings (1000, 2000, 3000 words etc.), and the program will color words above that level (that aren’t already saved) as… purple… for example. (We used to grey them out actually.) Maybe it could be a bit more smart, and would auto-adjust the level so that a certain percent of words are ‘white’ (‘suggested to learn’) for each video. Haven’t really though about it seriously yet, I’ll make a new thread for that discussion.

1 Like

Hey. Filipino fix should be online in next couple of days.

Sorry for the late response. I didn’t add the support for the language code in the title, but will shortly. It’s not boring information at, it’s a really nice project.

The typing exercise, ok, I will keep this in mind. Personally, writing exercises a not a priority for me, as I realised I never write things appart from in English. If I’m late crossing the border out of Ukraine etc. and I need to write a ‘protocol’ with the police, they help me with my spelling. :slight_smile: Most of our users are Koreans, Turks, Taiwanese etc. learning English though, and it might be helpful for them.

Keeping a database of the words the student has been able to type without mistakes could be a fantastic way to predict which shows, episodes, or videos are “comprehensible” for you.

This is an interesting feature for Youtube, but we don’t have a database of videos/subtitles currently. Coming soon.

I can expose some of the stuff we use for word frequency, getting lemma etc., as a simple command-line tool, it might be helpful for your project.

1 Like

Sure, we can add these, what language pairs (and directions) are they good for?

Btw, you guys had a recent mention in a big Youtuber account of a German language teacher that I follow. She is a big deal.

Oh, nice.

There is a conflict in creating Youtube playlists while using the extension, because of keys reassignments by the extension, which prevents you from naming them properly.

Added to ‘known issues’

I’ll suggest adding a similar overall ‘watch time’ tracking to the extension. Tracking all the videos you ever watched, of course, would also be an awesome feature (reviewing old content you studied).

Interesting. Adding to ‘Considering.’

Esperanto

Added to ‘known issues’.

Font size preference is currently a global setting, which means that I keep needing to switch it back and forth. It would be convenient to save a different font size preference, per different script, or per language, and to adjust the default values as well for each new script for optimal viewing.

Added to ‘Considering’

There isn’t enough space for some words to display fully, in the sidebar, at the bottom of the line. (Arabic)

Added to known issues.

This issue ain’t unique to youtube though… I’m pretty sure it’s happening on Netflix too.

Morfix is a Hebrew —> English as well as the reverse order dictionary.
Milog is a Hebrew —> Hebrew monolingual dictionary
And Pealim is like a special verbs dictionary that gives translation in English, extracts the root and provides conjunction tables.

2 Likes