 |
 |
On 10/2/12 8:44 AM, Paul Orkiszewski wrote: > Hi 4libers, > > Does anyone know of something - a kiosk, an iPad app, a web application > - that: I don't know of anything like it out there, but let's look at what it might take. I've done some software work in connection with Harvard's Iranian Oral History Project. > - Initiates an oral history interview by getting demographic info and > permission to use and stream for scholarly purposes. I'm not sure what you're saying here. It sounds as if you're talking about automated correspondence with the sources. That would be a huge project in itself, so I assume you've got something more narrowly focused in mind. > - Goes through a standard set of questions (in our case stuff about the > Appalachian State experience) There are two pieces to this: Recording the responses and storing the relevant metadata. The recording probably shouldn't be tied to a specific device or application, since field work can involve a lot of different conditions. The researcher in the field would want something to enter the metadata (who, what, when, where); this would be a straightforward piece. > - Stores the metadata, permissions release, and pointers to the audio > files created for each question in a dbase record You don't say what the scope of the work is; from the way you're putting the questions, I'm assuming it's a small-scale project with one researcher doing the interviews and putting the information together. Even so, It's probably best to have the field work be a separate application from assembling the information in the database. If nothing else, once you're at this point there's more standard software that can be used. > - Processes the audio through speech recognition either in real time or > post-interview, and populates the dbase record with rendered text (at > whatever level of accuracy) You could do this piece with Dragon; see this post for some discussion: http://www.nuance.com/dragon/transcription-solutions/index.htm A friend of mine is an expert in this area and might be able to answer some questions. > - Provide a search interface, where the meatadata, demographic info > (within reasonable privacy limits), and the transcript (however garbled) > is searchable. I'd suggest basing something on Apache Lucene. > - Crowd source the improvement of the transcriptions over time This needs to be better specified. One solution is to put the text onto a wiki. If you're talking about integrating it into the application that does all the rest, it could get messy. > - Package the interface as an app, and set up a machine image on Amazon > EC2, such that when someone uses the image and points a browser to it, > it goes through a set up routine so that smaller schools and historical > societies can set up their own sites in the cloud. I haven't tried > streaming on a free tier EC2 server, but you get 30 GB of storage, so > you could get a fair number of hours of audio (depending on the > settings) before you have to start paying. This, I assume, is why you're talking about treating the whole thing as a single application. Putting it all together would be a huge chunk of work. Dragon's software isn't free, and I don't know of anything for free that does decent speech transcription, so that would be a stumbling block to making it available to other institutions. > > ? > > Anyone interested in trying it with me if there's nothing already out > there? I'm leaning toward iPad, so we'd need iOS, server admin, dbase, > and media expertise. I have newbie-but-getting-better skill in the last > 3. Zero skill in iOS. I'm available for freelance work and it sounds very interesting, but you've just outlined a huge project that would be a significant burden even for the LoC's resources. That's not to say it can't be useful as a blue-sky starting point for something more reasonable. If you have funding, let's talk off-list. If you just want to continue blue-skying the idea for a while, I'm glad to continue on-list (and I promise not to bill you for that :). -- Gary McGath, Professional Software Developer [log in to unmask]