Thursday, November 12, 2015 1:06 PM
Many institutions have videos in their collections, but searching these videos for specific content can be challenging and time-consuming for staff and users alike. You can catalogue the contents of videos so that they are more searchable, but that can take a lot of staff time.
That’s why YouTube’s automated transcription feature is potentially helpful for institutions to have in their repertoire of tools. With a simple click of a button, a timecode synced transcript can be created for videos that you have uploaded to YouTube. While the transcript will definitely have errors, it gives you a great starting point. After editing the transcript can be added to the video’s database record giving your users a way to search the video’s actual contents.
When videos with speech are uploaded to YouTube, closed captions or subtitles are automatically generated. You can see if they are available by looking for the CC button in the bottom right. If the quality of the audio is reasonable, YouTube will also automatically make a text transcript which is accessible under the More hyperlink. This transcription can be copied and pasted into a field in your DB/TextWorks database and corrected while listening along to the video. This text can then be made fully searchable by your WebPublisher PRO or Andornot Discovery Interface software.
Even if you don’t want the videos to be permanently or publicly available on YouTube, you can still take advantage of YouTube’s transcript option. Videos can be uploaded privately, the transcripts generated and then copied and saved elsewhere, before the videos are then removed.
The accuracy of the transcript will depend on several factors, including the quality of the audio, the clarity of the speech and whether the subject matter contains many proper names. Apparently Scottish accents in particular can be problematic, but you can easily review the text first to check the quality.
As always, contact us for assistance to discuss possibilities for adding video transcriptions to your databases and search engines.