The Zombie Stack Exchanges That Just Won't Die
I know of many projects crowdsourcing the transcription of written material (manuscripts, letters, diaries, etc), but I don't know of any that are crowdsourcing audio or video transcription. Do any such projects exist? Better yet, do any such projects exist that have disclosed their software stack?
dsalo
I don't know of one specifically on crowd sourcing transcription, but at the 2010 ASIS&T meeting, Gary Geisler presented on Crowdsourcing the Indexing of Film and Television Media.
I seem to recall they were looking at stuff like where scene breaks were, what characters were in each scene, characteristics of each scene and stuff like that.
I don't know that it was true 'crowdsourcing', and I don't see any mention of a site that they created to do the indexing, but they do mention software called MovieBrowser (for which they reference Ali, N. M. & Smeaton, A. F. (2009). They also mention they used a modified version of GLIFOS
...
For large scale crowdsourcing in general, at the 2010 Digital Curation Conference, Chris Lintott gave the opening plenary, and in it he mentioned that Zooniverse had lots of people, but needed more projects for them to do. I know they're more focused on science, but you might want to contact them to see if there were parts that you could use of their infrastructure.
I'd also consider contacting the folks at reCAPTCHA to see where they get the source material for their audio CAPTCHAs. As they started out doing book OCR, I'd assume they'd consider helping with similar efforts in audio/video.
Have a look at http://pybossa.com/ it is not a transcribing project but can be used as the platform for making one, from the About:
"a free, open-source, platform for creating and running crowd-sourcing applications that utilise online assistance in performing tasks that require human cognition, knowledge or intelligence such as image classification, transcription, geocoding and more!"
Yes, they do exist.
So things like dotsub have been doing subtitling of videos for quite sometime and could basicly be used as is. The .sub format itself is a really simple way to keep timestamped text associated with an audio or video file and these guys have a simple tool to do that. There is also Soundcloud which has time based annotation features that could be used for this kind of thing.
There is also the Oral History Metadata Synchronizer. It is a really cool open source project that lets you sync transcriptions with timestamped moments. They are currently working out the details to build in crowdsourcing to the tool, so in that sense it doesn't exist yet. That said, the folks working on the project are claiming that functionality will come later this year.
Asside from this, their is at least one Zooniverse project that is a kind of audio transcription. It has the coolest name for anything around, WhaleFM, and they have you identify and group like sounds. I believe that all the Zooniverse projects are Ruby on Rails. A lot of their source code is up on their github page. If you wanted to see their audio stuff I bet they would polish it up and share that too.
You might also be interested in this post on visualizing oral history that is full of links about some of the neat work going on in oral history in this area.
Lastly, while it is not happening yet, I believe that the Storycorp folks are looking to get some crowdsourcing transcription stuff up and running as well.
I have not used http://www.universalsubtitles.org/ but they come up frequently in the open source community for subtitling and translating conference presentations. The project is completely open as well: https://github.com/pculture/unisubs/wiki/Dev-Center
What you're looking for is Amara. Its received an enormous amount of funding by both the Mozilla Fondation and the John S. and James L. Knight Foundation. Its source is all on GitHub.