(2015-04-19, 01:40)nickr Wrote: But would presumably add quite some size the the database. Also you're not going to want to index common words are you?
Also do you really want to index foreign languages? Some of my movies have subtitles in 10 or more languages.
Some subtitle formats are graphical not text, so not indexable or searchable without OCR.
Nice idea, but needs some thinking through I think.
PS I find Google assists me - search terms like "what's the movie where Arnie says I'll be back" are easier on the computer on the desk which has a keyboard.
Yes it would add a little bit of size to the database, but I think the performance hit is minimal since the database is not for an online website with lots of concurent requests and databases are pretty efficient at storing text
I was thinking to store the entire subtitle in a field like mysql's TEXT datatype, not to index words
For foreign languages I would say yes since i want to search both in my language and english or at least specify the languages I'm interested for searching
For the blu-ray's bullshit format I have no idea
I think it is like audio, it cannot be searched, at least for now
These two need additional libraries
Speech-to-text or voice recognition for audio
OCR for graphical subtitles
Not only that I don't know how accurate these would be, if they exist,
But it would also require a lot of power mking this unsuitable for devices like raspberry pi
Things like Google assist are nice but I don't want to be required to have an internet conection for this
(2015-04-19, 02:58)Ned Scott Wrote: I imagine the keyword search would be best to do via an online database/website, and not bother with an offline database of keywords.
One could potentially base it on some kind of "intelligent" (for a lack of better words) google search result, removing the need for any real database or community maintenance. Maybe like a bunch of site specific google searches (reviews sites, IMDB, themovieDB, wikiquote, etc) that give back URLs that can reliably be linked to specific movies, in order to cut down on false positives, and match that to a list of titles from your own database.
I'm more excited about the idea of doing a text search. Even if it's a painfully slow search that examines each video one at a time, or only one at a time, and only for text subs, it would still be awesome. I could point it to a recorded news program with a CC or sub track and look for keywords, and jump right to that point. Your own personal Snap Stream: http://snapstream.com
Or if you just want to quickly jump to a specific point in a movie, so you load up Scarface and search "Say hello to my little friend" and play that scene.
I LOVE this idea
I want that my database to be fully functionable without internet connection
If it can be improved or show results faster searching also on internet is ok but i don't want that internet connection to be a hard requirement
(2015-04-20, 21:33)lonjemoco Wrote: Not sure if related, but...
How about keeping (node-dependent) history of the search box? Sometimes it's annoying to type recent strings in the search box all over again, especially if you're not using a full keyboard.
That would be great
Also if you make a typo and you need to repeat the almost the same string like your last time with the option to edit it first.