Jarvik7 wrote:

It sounds like much more trouble than it's worth.

You're probably right. Or rather, not enough trouble! Must cut them up in batches after marking down the timings of sentence beginning/ends, and then rename the cut up files where the names are the sentences, and then rearrange them into paragraphs!

Last edited by ruiner (2010 March 20, 4:42 am)

I see...

Short chunks such as: http://speaking-japanese.com/breaking/index.html

I'm sure there are better, simpler ways of going about it. Just creating html links, for example. ;p I've just been obsessed with the idea ever since seeing TED's interactive video transcripts (for multiple languages). Something about having longform texts or sections of it smoothly integrated like that.

I guess in that sense YouTube can do the same, actually--with those t=0m00s addenda to the usual YouTube links... I've never made a YouTube video--do they allow one to use html links in the little sidebar/summary section? But then I guess you can just embed them anyway elsewhere.

Last edited by ruiner (2010 March 20, 1:03 pm)

Zarxrax wrote:

What are you trying to accomplish that you think you need cue sheets for video files? There is probably another way of doing whatever you want to do.

Interactive video transcripts (these are actually translated into Japanese many times on TED, just select from the dropdown); controllable audio accompaniment to non-SRS reading of paragraphs or longform texts without substantively breaking the flow of the reading experience to activate said control, with the means to use hotkeys perhaps or individual sentences to obviate the need to then deactivate or stop the sentence. Perhaps there's a way to actually generate those #t=00m00s links or .cue sheet type sections per click, so that one can easily select, backloop, etc...

Last edited by ruiner (2010 March 20, 2:07 pm)

Zarxrax wrote:

So basically, something just like sub2srs that just leaves everything as 1 video file, and lets you jump around between sentences?

I just noticed ahibba's comment in another thread. Subtitle Workshop is free and allows you to load and preview the video/subtitles while editing, so you can just click on the lines below the embedded video preview and it takes you to that spot in the video. So I guess that does work for non-SRSing of (raw) videos. Somehow it feels unsatisfying, but I'm not sure why yet, will have to tinker and brainstorm.*

Now for uninterrupted reading w/ controllable audio...

*Oh right, remembered the nascent strategy that spurred the idea in the first place... (half-formed thoughts, bleh)... it's because I was thinking of it as easily shared/streaming video, where the idea is to have these lightweight texts that are easily skimmed and searched for interesting content and then not having to download an entire video or having tonnes of videos stored in a one-to-one relationship, or having to DL and convert them into clips via subs2srs...

... and also (for that first step of looking at the idea of converting stuff that's already on your HDD) I was thinking of having subtitles kept separate from the video when playing... I find English subtitles for English distracting and English for Japanese as well, and imagine that as one progresses, the same could be said for Japanese. This would follow the redundancy effect sometimes found in studies of multimedia learning which causes less information to be retained with more effort in video w/ audio narration and on-screen text all at the same time, but other studies showed that's when it's less important to use the orthography to cue phonetic awareness or vice versa, and more about speed of narrative and semantic content.

I guess subs2srs clips can work in Anki--doesn't the new version of Anki allow media previewing from the card browser? Hmm. I feel somehow like I've wasted everyone's time here. ;p Should've just stayed focused on the audiobook thing and not even mentioned video bookmarks.

Last edited by ruiner (2010 March 20, 2:41 pm)

Zarxrax wrote:

Going back to the original post, it seems to me that the obvious solution is to mux the video into a matroska container with a chapter list.
It would be fairly trivial to write an application to generate a chapter list from a subtitle file, but I don't have time to do it now.

That would involve a relatively unwieldy processing of the video, no? Then I'd think of that as 'more trouble than its worth', as J7 would say. ;p (See above asterisked edit.)

Last edited by ruiner (2010 March 20, 2:37 pm)

So wait, there's no way to preview the media in Anki cards? Weird. (Weird as in I think I hallucinated that this was a new option in latest version.)

Last edited by ruiner (2010 March 21, 10:41 pm)

Zarxrax wrote:

...and all chapters are available from a rightclick menu. Do you need even more than that?

I'm not too fond of using a context menu. Think about a 24 minute anime with several hundred sentences. How many chapters do you think would fit on the context menu?

balloonguy wrote:

If I'm understanding you correctly, here's a quick proof-of-concept that uses HTML5's audio tag. This will work in Firefox 3.5, Chrome 3 and up, and in any other browser that supports the audio tag. Clicking on a line seeks to that spot of the audio. This can work the same way using video.

Oooh, that's precisely what I mean. It even uses dropbox! So you can distribute a bit more easily...

As I mentioned above, I'm kind of into the idea of not only sharing (and prioritizing the subtitle/text information so you can decide whether to take the trouble to use bandwidth/storage based on what's in the media file), but keeping the subtitles off-screen, as it were, for a semi-raw experience. Perhaps have a parallel text set up if there's English subs. If we distribute like that we could even add character dialogue tags like a script...

Likewise with audio books, I want to do something like controlled shadowing, where you can easily 'backloop' at your own pace. I can see going through an audio file and marking down when the first word in a sentence appears (depending on length of sentence/phrase), and then adding that timing information and integrating it into the text as a link to 'seek' with. Oooh, and even adding links to the ruby for WWWJDIC audio? And maybe at the end have a link to Anki deck of vocabulary? hehe, Getting ahead of myself...

I guess an audio-only version of the former is more economical and sort of like a mix with the latter. There's also the possibility of snapshots img src'd into the text...

We have links elsewhere for public domain audio/ebooks, plus there's Reading Real Japanese and Breaking into Japanese Literature and similar resources for the pirates among us, who are very bad and I disavow any knowledge of.

Last edited by ruiner (2010 March 21, 10:36 pm)

balloonguy wrote:

Here's a concept about keeping the subtitles off-screen for video. You can still seek by clicking on the text. However, on the side, if you need to check what was said, clicking the check text link displays the previous line, the current line when clicked, and the proceeding line.

If your in the US, here's the same thing using episode 39 of Full Metal Alchemist Brotherhood. This does not work in Firefox, only Chrome and Safari.

Nice! One problem seems to be that scrolling down to later areas of text meant missing the video. ;p Maybe a separate frame/scrollbar thingy for that would be better? Load time was also slow and playback a bit glitchy until it worked itself out, so I guess audio-only has a plus there. Could've just been server load or maybe just me.

So if we had a 'tool' or something for this, in addition to sharing subs2srs decks, we could share streaming video with (searchable) interactive transcripts! Amazing... ;p Or use it as an offline player for whatever totally legal videos we have? Not just for Japanese subtitles, but as we see, for English when people want to make a transition to raw but also wouldn't mind quick English subtitle support (I know I hated deciding whether to start missing out on my favourite shows or risk spoilers/repeat viewings with and without subs).

I guess if you're making markers yourself for an audio file or something, you'd have to stick it in a certain format before converting to whatever that html stuff is?

Last edited by ruiner (2010 March 22, 9:11 pm)

If you could make it a URL or Browse for video/audio, that would be cool. ;p So we could use it standalone or in conjunction with dropbox or whatever... Perhaps share links on another site to avoid problems here (in case someone accidentally linked to a copyrighted file). Would the structured text have to have an 'end' point like a subtitle file, or can that be ignored. (I haven't figured out a good way to make timing markers for audio/ebook combinations to do 'controlled shadowing', so.)

You are awesome, thanks for playing around with this. No one else is commenting so perhaps we're just crazy, but I think we're on to something!!

I was making a few notes earlier, forgot where, but I was thinking about the possibility of embedding YouTube/streaming video from other sites (in case dropbox or wherever proved problematic as video host, though obviously the video/subs would still need to match)--no idea how that would work in this case, what kind of URL format is required...

Last edited by ruiner (2010 March 23, 11:12 pm)

balloonguy wrote:

The biggest problem for video is that Firefox only supports Theora codec videos, Chrome supports Theora and H.264, but Safari only supports H.264. So, if people wanted to use their own videos, they may have to re-encode them, because for me most of my files are not H.264. However, as demonstrated, Youtube videos work fine in Chrome and Safari, and I think most video sharing sites use H.264 anyway. So, one would just have to figure out the url of the video. According to this, there's no set bandwidth limit for sharing using dropbox, but they will throttle if it gets excessive. 

Chrome is on track to allow loading local files from a remote website in version 5, but all modern browsers allow reading a local file if the webpage is also local. Also, an end point is not required for this.

Yeah, hopefully the h.264 thing won't be a big deal. So how would I, say, load a local audio or video file into that URL offline? So what kind of format does the conversion require for text?

For example: If I had audio for this paragraph and marked down certain times:

"[00m00s]Yeah, hopefully the h.264 thing won't be a big deal. [00m0xs]So how would I, say, load a local audio or video file into that URL offline? [00m0xs]So what kind of format does the conversion require for text?" Would that work? ;p </eager to experiment with an audio/ebook>

balloonguy wrote:

This allows you to specify both the audio/video file and subtitles. You have to manually enter in the url, so if the file is locally on your machine the url will be, for example, "C:\users\bob\vids\vid.ogv". Also, if you are not on Firefox, the subtitle input box will change from allowing you to browse your files to making you manually specify the url the same way as you have to for video, but the webpage must be on your machine. I added ASS support, but will work on adding the format you specified.

Too cool!! Thanks. I can decide which I'm more excited about, the video aspect or audio. I guess audio at the moment because we don't have much in the way of tools to tinker around with that, re: reading/audio or text+multimedia outside Anki.

The video thing alone and the use of dropbox + html + transcripts, etc., I think allows for all sorts of stuff related to sharing materials beyond Anki. That's something we haven't really touched on before, it's all been decks and spreadsheets...

Edit: Hmm, I'm actually unable to load either an .srt or a video file now. ;p\

Edit 2: It seems the subs you linked work, but not the others I have. The difference seems to be yours had the "# [timing]" on the same lines, whereas most of mine don't seem to. The URL video loading works, but not typing in the local path in quotes or without. I also tried changing the format a bit so instead of c:\ it would be c// but that didn't seem to work either. .ass worked, though.

Last edited by ruiner (2010 March 24, 11:30 pm)