Audio Visual Oral History

Viewing 6 reply threads
  • Author
    Posts
    • #132263

      Does anyone have a voice recognition tool that will search AV and/or audio oral histories for keywords? From what I understand organizations such as StoryCorp are still manually listening to and flagging every few minutes in order to create keyword indexes, but I am hoping that current technology is exploring software options that make it possible to automate the indexing process. Any ideas or suggestions from the community? Thank you.

    • #132269
      Michael Hosking
      Participant

      While I cannot answer this specific question, there is a website created “Oral History in the Digital Age” (http://ohda.matrix.msu.edu/).  I did a quick search and came back with a few results.  There is a contact link and it might be worth contacting the creators of this site and see what kind of response you get back.  I searched “Tagging” and came back with some results that might help you find and answer.  I hope this helps and sorry I could not provide you with more.

    • #132268

      I am an oral historian with a research library and I have not heard of such a break through as a voice recognition tool that can search for key words.  The oral history community have been poised and waiting for voice recognition software that is capable of transcribing more than one voice at a time, so that transcription can be partially automated for interviews, but even the sophisticated systems now available are not quite there yet.  As far as I know tagging is still a manual business, but I am sure in the future this kind of software will become commonplace and I agree that the website “Oral History in the Digital Age” is certainly the right place to look for its arrival.

    • #132267
      Kaia Landon
      Participant

      No, that future is not here yet.

      While there is software that can transcribe audio (Dragon Naturally Speaking is the main one), it relies on being trained to a specific speaker. You speak into the software and correct it when it gets things wrong, so that eventually it learns how you speak, and transcribes accordingly. Since it has to be trained, there’s no way to use it with pre-recorded audio or video.

    • #132266

      Try looking at Doug Boyd’s blog at The Louie B. Nunn Center for Oral History at the University of Kentucky: http://libraries.uky.edu/libpage.php?lweb_id=11&llib_id=13

      Boyd has developed the software package OHMS, which allows organizations to create a digital database specifically for oral histories. It publishes the history on line, and lets you link abstracts, indices, or transcriptions to the audio or video. It allows for keyword searches of the transcription that will fast forward the video directly to the keyword. However, the interview MUST be transcribed and the time code of the video matched to the transcription. Very labor intensive. Figure four to eight hours of transcription work for every hour of video. And it is not something that just anyone can sit down and do.

    • #132265
      Michael Nagy
      Participant

      I think the future for access to large aural collections lies in speech analytics rather than voice to text.  Many large media companies are employing dialog searching to seek across videos, television programming, and news footage.  There is no transcript and they are not wanting to create a transcript.  Just searching the audio directly – kind of like OCR for voice.  In this scenario you would not create a transcript as a search mechanism but may only create a transcript or partial transcript on an as needed basis for documentation as text, such as quotations from the speaker.

    • #132264

      Michael: Thank you for your thoughts.   I look forward to a continuation of this thread once I have a chance to investigate.

Viewing 6 reply threads
  • The forum ‘C2C Community Archives – 2012 through 2014’ is closed to new topics and replies.