A method comprises receiving video content from at least one video source, timestamping frames of the video content to generate respective first timestamps, and providing the frames in association with the respective first timestamps to a commentator application. The method further includes receiving audio content from the commentator application in association with respective second timestamps, combining the frames of the video content with the received audio content based at least in part on the first timestamps and the second timestamps to generate commented video content, and providing the commented video content to one or more servers of a content delivery network for delivery to one or more viewer devices. A plurality of distinct streams of audio content may be received from respective distinct commentator applications, and the frames of the video content combined with the plurality of distinct streams of audio content to generate the commented video content.