Why AI-driven broadcast can be a game-changer

October 12, 2017

By Harshad

Machine Learning or AI in real life works far more differently than science fiction would have you believe. The fundamental question we need to ask ourselves is whether we need an artificial intelligence at all, and what for? As far as the broadcast industry is concerned, the answer is quite simple. With increased globalization, and higher adoption of OTT video, we are witnessing a content explosion, and Machine Learning could address many issues caused by it.

Content is the new bacon

As internet became omnipresent in early 21st century, potential pipeline for content production widened exponentially, and irreversibly. TV networks eyeing to expand into new territories, or new content consumption platforms are facing a curious problem. Unlike individual content owners, TV networks need to maintain flawless quality of their content. Segmenting thousands of hours’ worth content manually can multiply costs, and not doing so could push the audience away. With Machine-learning it could not only be possible to create logical video segments, but also perform QC for aspects such as freeze frames, noise, black frames etc.

Live to VOD via deep neural networks

A decade ago streaming live was a forgettable experience. Mainstream TV was unchallenged when it came to quality live experience. While this still holds true, parallel content consumption is on a steady rise. Consumers today want to move seamlessly between devices and may want to catch specific parts of live events-almost as soon as they are broadcasted live! Creating VOD segments manually can be challenging and inaccurate. On the other hand, with Machine Learning it’s possible to generate OTT-ready segments in an instant. The machines could be taught exactly when the live reception starts, when it stops, and to remove any ad breaks in between.

More money from OTT with automated ad detection

Today, ad detection is largely an ad-marker driven enterprise. However, TV ad spots cannot be ported into OTT space, mainly because attention spans and other user behaviours differ greatly between the platforms. Manually detecting ads, inserting ad markers, and communicating the same to various content delivery platforms could require setting up additional teams. Machine learning could do this cost-effectively by using intelligent ad mining engines. What’s more, as more channels start using AI-driven detection, the accuracy of such services will further improve.

In many ways, broadcast and advertising is set for the next stage of evolution, and companies like Amagi are at the forefront of it. Reach out to us to know more about our innovations in the space of machine-learning and AI!