NHK recently launched an auto-generated program script system in its archive database. Files of news and other programs are transmitted from the play-out server to the archive server after broadcast. At the same time, all the metadata (including copyright information) is sent to the archive information system.
The auto-generated program script system automatically generates a program outline table containing descriptions of sequences and shots, narration scripts, subtitles, copyright information, and conditions of use for each program. This system can recognize the breaks of each shot automatically, so it places an image of the first frame of each shot in the thumbnail column. It takes narration scripts from a closed-captioning service.
The outline table is very useful for producers as they can easily grasp the content of each program by browsing. By clicking a thumbnail, users can watch a reference video. By clicking a checkbox, users can request a copy of the sequences or shots. Users can also access detailed copyright information (for instance, payment information, PDF files of contracts, and trouble reports) on each shot or sequence.
We are also pursuing functionalities of automatic description for the database: content-based image retrieval which enables users to search target shots based on similar shots without using keywords, auto-generated video segmentation, and automatic image tagging which helps reduce the burden of manual work for cataloguing. STT (Speech to text) is also used to generate news scripts from newscasters’ voices.
In addition, we have created a website “NHK Chronicle" where users can access past program guides and descriptive metadata dating back to 1953 when NHK started its TV broadcast. On this website, users can search programs they wish to watch in the VOD service or at NHK Open Program Libraries where people can watch past programs free of charge.
This complete metadata management will simplify the procedure for reuse of content for productions, online services, and distribution. NHK hopes that it will promote utilization of stored programs and eventually contribute to reducing overall costs including location expenses.