ruiner
Member
Registered: 2009-08-20
Posts: 751
I thought this might get more attention if I posted a specific thread about it, but balloonguy is making something cool here: http://forum.koohii.com/viewtopic.php?pid=98087#p98087
It's still a work in progress we (as in just balloonguy) are working the kinks out for, but I think it's going to be something very cool when it's done!
Basically it can allow us to share videos and Japanese subs at dropbox or wherever, streaming videos there, and being able to click on any line to get the video to play at that point--which means you can watch the video raw, but have the interactive transcripts to follow along, and if you miss something you can just click and scan back. I think it could help for when you want to transition to raw but want more control of when to use even the English subs as well. I explain my rationale more in that thread.
For me the big thing is 'controlled shadowing' -- loading up an audio book and the ebook (for instance, the public domain works posted in the stickied thread on this forum (also on our wiki, Audiobooks, or your own legitimately purchased offline books/readers w/ accompanying audio that you totally shouldn't share with others on dropbox) and being able to click any sentence to scan to the audio for it while you're reading, or when you're shadowing, to loop back to it as you like (ever since reading Ong's Orality and Literacy years ago, I call it 'backlooping'). I think that kind of reading practice outside Anki could be very useful.
Balloonguy is still working on the format/script aspect of that, but the idea is that a person or persons would listen to the audio and just mark down when a sentence begins, or whatever, depending on the desired length? And end markers aren't necessary. Then just stick that in the ebook and voila. I guess clever people could think of ways to streamline that process, but once it's done, that's all you need to share, besides the audio which can be streamed from a URL or something.
At any rate, perhaps this is all a bunch of pointless nonsense, but for some reason I'm really excited about it! ;p
More on 'backward scanning' and 'backlooping' -
http://books.google.com/books?id=v21leP … mp;f=false
http://books.google.com/books?id=v21leP … mp;f=false
Last edited by ruiner (2010 March 25, 1:49 am)
Looks like that might work, thanks! I'll check it out. Damn, Chinese sounds good...
Edit: The key thing for me is to be able to click the text--formatted just as how it is as an ebook--and seek to that area in the audio. This tool might not be able to do that, but it could be good to create the texts to import into balloonguy's tool.
Last edited by nest0r (2010 March 27, 6:09 pm)
ahibba wrote:
nest0r wrote:
The key thing for me is to be able to click the text--formatted just as how it is as an ebook--and seek to that area in the audio.
Like this one?
That's amazing! It's even better than what I was talking about, as it highlights the text as it proceeds and has multiple translations that pop-up. What's funny is that I acquired that Petit Prince audio/ebook a couple of years ago for future French language study. Seemed like it was good for beginners.
I couldn't find this stuff on the person's homepage, and the download areas of other sites explain it as being made with a special "audio-text synchronize" program. Likewise the installation folder is 'Audio-Text Synchronized Books'. I wonder where or how one might obtain such a tool, and how it could be used for other books. If there is no such tool available, we'll be best served using whatever balloonguy fixes up, methinks.
Edit: I think this is it?? http://www.interactiveselfstudy.com/
Haha, I see your first link was from that site. The problem remains, though, I don't see any way for making our own--I guess they and that taudio site want people to just let others make books for them or something? Seems like a pain to get their stuff working w/ a special reader when we're only a step away from doing it ourselves...
Last edited by nest0r (2010 March 27, 7:25 pm)
Looks like that slattery person went even further w/ using Transcriber, I'm still parsing the information though: http://librivox.org/forum/viewtopic.php … 8793d15496 + http://www.dinglabs.com/ <--This might be exactly what I want, once I figure it out and see if it's available offline... + http://blog.dinglabs.com/
@balloonguy: I think we can use Transcriber to make .trs files and import those if you can arrange that... that would allow for offline/dropbox use rather than having to use one reader at dinglabs (or perhaps there's other ways to use that, still looking into it): http://blog.dinglabs.com/2010/02/how-to … ntent.html
Mostly I'm concerned about investing effort in synchronizing texts, only to rely on a single person's website to actually use them all.
For instance, here is described the process of using your own synchronized books by inserting "#b:=http://book xml" into the URL. I wonder if one downloads the Reader from the browser, how to go about doing the same? Such as "file:///C:/Book.htm#b:etc"...
Last edited by nest0r (2010 March 27, 7:46 pm)
Slattery in the Librivox thread mentioned this (guess it's English only) forced aligner: http://www.ling.upenn.edu/phonetics/p2fa/ - I could've sworn I've seen something like that mentioned here already. This being part of the automation process, re: Transcriber. I guess a cheap, quick way is still to do something like find/replace like "。" with "。[timecode]" or use preset timespans or something... if that's the case then Transcriber I suppose wouldn't be necessary? Though perhaps there'd be a way to make the formats interrelate... hehe
Yes, clearly I should be working in some corporation, heading up project management for software development. Creating Rube Goldbergian programs to make everyone's life easier. The trick is to make the process itself into a Rube Goldberg machine.
Last edited by nest0r (2010 March 27, 8:37 pm)
@Axlen
I couldn't seem to get a blog comment to post. Could you ask them about standalone/offline stuff while you're over there? ;p Some way to not have to invest effort into creating all these synchronized texts and rely on a single website to read them... If they want to keep things as they are, it's not a big deal, I think balloonguy's tool is only a step away or something from being as good for audio/ebooks as for video + interactive transcripts.
It's funny, those interactiveselfstudy and taudio sites, they give the reader away for free, but you have to let them make the texts from public domain works. Here it's just the opposite, you make your own but have to use the site to read.
Last edited by nest0r (2010 March 29, 4:32 am)
Just an update, so far the latest version balloonguy added works well for me: http://forum.koohii.com/viewtopic.php?pid=98695#p98695 - For Firefox audio, I had to use Audacity to quickly export as ogg/wav.
What should we call this tool?? 影真似の術 (かげまねのじゅつ) aka Shadow Imitation Technique? 影分身の術 (かげぶんしんのじゅつ) aka Shadow Clone Technique? Shadow Skill? ;p 影縛り? /nerd
Here's how to use Transcriber: http://blog.dinglabs.com/2010/02/how-to … ntent.html
Again, here's many Japanese audiobooks w/ transcripts: http://rtkwiki.koohii.com/wiki/Audiobooks
Personally I think we could do whole books, but for now perhaps just swaths of text, or group projects of longer length texts would work*. Sharing them at the wiki as Axlen did seems like a good idea. If some rakes decide to work on legally delicate stuff such as Harry Potter, graded readers, Breaking into Japanese Literature, Read Real Japanese, etc., rather than the public domain linked above (there's also Librivox's site for I'm assuming only non-Japanese works), then I guess that'd be trickier to share/link to.
As for use for videos, I'm thinking those, being larger, could be more dependent on creating a kind of streaming video hub via dropbox or drop.io or some similar kinds of sites.
To repeat, my main idea for audio is 'controlled shadowing'/backlooping ability via sound files. For video, it's to move past using subtitles for listening, per the exception to the 'redundancy effect' in multimedia learning and simply focus on listening alone and following the meaning and focus on the actual video rather than subtitles (there's also that annoying uncanny valley effect as I think of it, with English subtitles). In other words, it's just a way to transition from any sort of on-screen subtitles but still get the benefit of searchable, interactive transcripts at hand, as well as having a nice means to share both text and video. I think I had other reasons but forgot as I'm focused on audio-/e-books.
*See http://forum.koohii.com/viewtopic.php?pid=93872#p93872 and http://forum.koohii.com/viewtopic.php?pid=69874 and maybe http://forum.koohii.com/viewtopic.php?id=4560 (this latter on 'lexical priming').
Last edited by nest0r (2010 March 31, 1:39 pm)
So I'm finding Transcriber really difficult to figure out. Any tips?
Nevermind, it's actually really really easy. Suffered a brief brain glitch, I guess. ;p
It's actually pretty fun once you get the hang of it. I think I'm doing something wrong though, the timings seem right when I'm doing it, but they're slightly out of synch when I play them back. But I think resynching isn't too hard, though I haven't tried yet.
Last edited by nest0r (2010 April 01, 12:08 am)
Axlen wrote:
nest0r wrote:
... I think I'm doing something wrong though, the timings seem right when I'm doing it, but they're slightly out of synch when I play them back. But I think resynching isn't too hard, though I haven't tried yet.
Per Jim at DingLabs Transcriber has trouble with long mp3 files. The solution is to always time using wav files. There are plenty of programs that can convert an mp3 to a wav file. For shadowed playback though you can still use the mp3 version of the audio.
Yes, unfortunately I was already doing that, having followed and noted above (alongside mention of Audacity for file conversion) your conversion on that blog, but there's still synch issues with Kage Shibari (that's what I'm calling it till someone makes fun of it or balloonguy decides on a name for his tool ;p). It's not a big deal because usually it's like ~250-500ms, and I'm sure there's an easy fix. This standalone tool is still in its infancy, after all! I'm already wondering about stuff like adding images, parallel text, paragraph #s, ruby markup, glossary with JDIC audio, etc. I mean it's ultimately an HTML file, after all. Or something. But that whitespace below the media player to the right keeps beckoning to me to add something... (Horror vacui.)
To be honest I'm not sure I'm too worried about Transcriber and mp3/wav files--I'm going to test out whether it has issues with 5-10 minute ogg files (can it even play those?).
Last edited by nest0r (2010 April 01, 2:55 pm)
So has anyone made use of this tool? Since the awesomeness of resources out there tend to pass people by (esp. when they're presented in scattershot, cryptic fashion as they're developed here on the forum), I'm thinking of doing a major project related to this, w/ instructions. That is to say, a collection of interactive audio stories + instructions on how to use balloonguy's tool to make them work.
Hopefully others will feel inspired to do the same... there's a lot of audio-/e-book stuff out there...
Example here: http://forum.koohii.com/viewtopic.php?pid=98779#p98779 (Make sure to grab updated .trs from second link.)
Last edited by nest0r (2010 October 23, 6:09 pm)
