Script metadata streamlines production

Many people in broadcast production would agree that metadata is essential to the smooth flow of program creation, distribution and delivery. But, there is a catch. Whoever records the metadata rarely receives the benefit; it is someone further down the workflow that reaps the rewards. So, there is little incentive to add metadata beyond what is necessary for the task in hand.

Consumer applications can already harvest metadata from smartphone location services and from picture analysis, albeit not always with great accuracy. Such automation of metadata harvesting is beginning to be adopted in professional products. This automatic collection of metadata certainly relieves creative personnel from what they may consider as unnecessary tagging.

Metadata has obvious application in the editing process, but it can also help downstream in program distribution. The pressure is on broadcasters to create many versions of their content, for different languages and for different devices, to meet demands of the global market. Foreign language versions involve audio dubbing or subtitling, both labor-intensive processes.

The process starts with a transcript. In the absence of a script, then it must be created from scratch, possibly by re-voicing, then using a speech-to-text converter on the re-voice audio track. Re-voicing creates a clean audio track with no background noise and an even level. It also means one speaker to learn for the conversion software.

More products are adding speech analysis, including non-linear editors and newsroom systems, so that editors and journalists can search video for specific words and phrases. But, no speech-to-text system is perfect, and results can often be less than useful. Sometimes, the original script exists, which can be of great assistance in improving a transcript’s quality. Even if all the words are not recognized, sufficient matches can be found to align the script with audio tracks.

That opens up the concept of navigating along the timeline from the script, rather than looking for cues in the video. Once the script is embedded as metadata linked to the timeline, it presents a good starting point for creating a new language version.

In searching for improved efficiency throughout broadcast operations, two good examples that exploit metadata are the use of camera-mounted GPS receivers for location data, and the use scripts downstream of acquisition. Both examples are good in that they do not place an extra burden on the operator. The term script-to-screen is truly becoming a reality for metadata.

Stories in this newsletter from Turner and ITV show how tools from Nexidia and Adobe, respectively, are helping broadcasters make their operations smarter. Notepads, Post-it notes, and videotape all have their place, but the intimate association of metadata with video and audio files allows producers and broadcasters to leverage the power of computers to streamline operations.