IBC 2018: Microsoft Azure’s AI-based Video Indexer metadata extraction tool now available
At this year’s IBC, Microsoft Azure (1.C27) has announced the general availability of Video Indexer, its AI-powered advanced metadata extraction service. Announced as public preview earlier this year, Video Indexer provides a set of cross-channel (audio, speech, and visual) learning models.
According to a blog post by Sudheer Sirivara, Partner Director, Azure Media and Azure CDN Services, Video Indexer is a prime example of Microsoft Azure’s Cloud + AI focus. It brings together the power of the cloud + Microsoft AI to intelligently analyze users’ media assets, extract insights, and add metadata. It makes it easier to understand vast content libraries and get the more than 20 new and improved models, easy to use interfaces, a single API, and simplified account management.
Announced as a public preview, Video Indexer orchestrates a cross-channel machine learning analysis (audio, speech, and vision) pipeline for video and audio files, using models that are continuously updated by Microsoft Research. These models bring the power of machine learning to the user, creating insights that can then be used to improve content discoverability and accessibility, create new monetization opportunities, and unlock data-driven experiences.
Customers and partners such as AVID, Ooyala, Dalet, Box, Endemol Shine Group, AVROTROS, and eMAM are already using the Video Indexer service for speech to text and closed captioning in ten different languages, visual text recognition (OCR), keywords extraction, label identifications, out of the box and custom brand detection, face identification, celebrity and custom face recognition, sentiment analysis, key frame detection and more.
The Emotion recognition model detects emotional moments in video and audio assets based on speech content and voice tonality. The Topic inferencing model is built to understand the high-level topics of the video or audio files based on spoken words and visual cues. Topics in this model are sourced from IPTC taxonomy among others to align to industry standards. Microsoft has also enhanced the celebrity recognition model, which now covers one million faces based on commonly requested data sources such as IMDB, Wikipedia, and top LinkedIn influencers.
CLICK HERE for more info on Video Indexer.