Video logging

Have you ever wondered how broadcasters find relevant content among the thousands of hours of daily produced video or in their archive with decades of material? Well, they use video logging methods to index as much content as they can. They search in content realtime.

In early days videos were actually watched and labelled according to their content by humans. An increasing number of broadcasters transforms this video logging workflow from a manual to a fully automated process. In this blog, I explain why this transformation gains momentum and more important how this transformation helps to increase output quality and quantity and cut costs at the same time. Machines are now able to recognize speech close to human recognition levels for clear audio. Tech companies such as Microsoft, IBM and Google made serious technological breakthroughs in this field with many companies in the industry in their slipstream for general ASR. Other companies in the field have focused on niches and outperform general ASR models by customizing on use-cases in medical and legal verticals. At Zoom Media we’ve trained our ASR on broadcast content outperforming other companies on this type of data. 

Increase the Quantity of processed videos
In this digital era, broadcasters produce and distribute content to reach consumers in an omnichannel way. Think of a reality show or a news channel with 24/7 coverage. All you see are the most important stories on TV and to continue watching you’re referred to theme-channels, OTT-platforms, online streams and social media. With the growing amounts of video in the production flow and more live content, it has become impractical to log each clip manually. Also, it’s time-consuming and expensive to find relevant clips during editing. Besides these obstacles, broadcasters are often required to retain records of all emitted content. To organize this process more efficiently we help broadcasters with automatic logging prior to editing and in real-time for live streams using our Zoom Media speech to text model for broadcast content. 

Enrichment in the Video Logging Process
Modern Video Logging doesn’t stop with transcribing the videos automatically. NLP technologies can, for example, be used to categorize content automatically. Broadcasters harness the power of these and other technologies in Microsoft’s’ Video Indexer where speech to text is combined with technologies such as visual object detection, OCR, face recognition, named entity recognition and many more.

If you would like to discuss possibilities to elevate your logging process or just want to learn more about our technology you can contact me at ab@zoommedia.ai 

Share this post

Share on twitter
Share on linkedin
Share on email