AUTOMATED POST-PRODUCTION EDITING FOR USER-GENERATED MULTIMEDIA CONTENTS

摘要

Methods, apparatus and systems related to packaging a multimedia content for distribution are described. In one example aspect, a method for performing post-production editing includes receiving one or more footages of an event from at least one user. The method includes constructing, based on information about the event, a script to indicate a structure of multiple temporal units of the one or more footages, and extracting semantic meaning from the one or more footages based on a multimodal analysis comprising at least an audio analysis and a video analysis. The method also includes adding editing instructions to the script based on the structure of the multiple temporal units and the semantic meaning extracted from the one or more footages and performing editing operations based on the editing instructions to generate an edited multimedia content based on the one or more footages.