IT Brief New Zealand - Technology news for CIOs & IT decision-makers
Story image
Isentia infuses AI into media monitoring capabilities
Fri, 5th Feb 2021
FYI, this story is more than a year old

Media monitoring firm Isentia has infused its services with a dose of artificial intelligence in order to better detect the beginning and end of relevant mentions in broadcast media such as press conferences, television, and radio.

Isentia's new feature, called Boundary Detection, is designed to isolate relevant news items that would otherwise be difficult to determine in some forms of media.

The company developed its artificial intelligence (AI) model by using thousands of hours of news programmes that were segmented manually. Now with AI, detection is an easier process - and it's also a ‘milestone' for Isentia.

 “The new functionalities, which leverage cutting edge machine learning and business logic, are testament to Isentia's dedication to putting customers' needs at the forefront. We are committed to ongoing innovation and we pride ourselves in providing a sophisticated media intelligence service to our customers,” comments Isential chief executive officer Ed Harrison.

Isentia chief technology officer Paul Russell says that news stories can come in different forms. In some cases it's easy to detect the beginning and end of relevant stories, it's harder to find them in press conferences or talkback radio.

“Isentia's proprietary technology is able to automatically detect the beginning and end of any type of news item, even those that involve a variety of topics, or several speakers,” he explains.

“We've also upgraded our media player to give communication professionals full control of video and radio content so they can quickly jump to the relevant keyword or watch it again.

Isentia has also built a new way to filter out different forms of audio such as news, advertisements, and music. This technology also uses AI to detect different sound waves. That means that ads and music can be removed from broadcast monitoring.

“Human input has helped develop the AI model by spending months monitoring and annotating news data across TV and radio. We then trained the model to mimic the professional annotators and recognise which content was advertising and which content was a news broadcast,” says Russell.

"In the same way we can tell the difference between news and advertisements if we listen to a foreign language program, our technology is able to detect those differences and ensure our customers only get what they need in their media coverage, without the interruption of ads."

Isentia states that the new AI model behind the ad filtering features will learn and evolve based on feedback from both client managers and customers.