Panopto Greatly Enhances Video Search Capabilities in Version 4.6
Let’s face it: while video in the enterprise is good for many things, search is not one of them. That one-hour video you have of the product manager explaining the competitive strengths of the next company-saving product is just an amorphous opaque blob of information without a lot manual tagging and metadata insertion. And the only thing people like less than watching a one-hour amorphous opaque blob is manual tagging and metadata insertion.
Many companies in the lecture capture/webcasting space make their videos searchable by automatically ingesting text from PowerPoint or Keynote slides used in the presentation. The major focus of Panopto’s (that's the company and product name) 4.6 release is to take these search capabilities much, much further.
Specifically, Panapto added automated speech recognition (ASR) and optical character recognition (OCR) to its system as sources of searchable metadata. The application of the first is obvious; any speech in the video is automatically transcribed and associated with its timecode. Those searching the presentation for transcribed words will get links to those spots in the video (Figure 1).
Figure 1. Mentions of Android at the Google IO Keynote.
According to Panopto, processing time for ASR is about 1X the video length, so one-hour of video will take one hour to process. The company did not share any details about the accuracy of the transcription, but during the demo the software pulled the name Bill Gates out of a Gartner presentation, as well as several other names and acronyms such as ROI.
The second new feature, optical character recognition, recognizes text contained in video frames, like text in graphics in a PowerPoint slide, or video of a whiteboard (Figure 2) or other objects with text. By default, the system samples once every ten seconds with a processing time of .25X the video length.
Figure 2. Grabbing text from whiteboard video and graphics within a PowerPoint slide.
Again, no word as to accuracy, but during the demo, the system was able to decipher the product number from the tag on a laser printer cartridge, which was impressive. While not the first to add either OCR or ASR to its products, Panopto claims to be the first with OCR, ASR and text injection from slides and speakers’ notes. Whether first or not, if you’re currently using Panopto, it’s a great bump in functionality, and if you’re in the investigation or buying phase, it may be a useful advantage.
In the 4.6 release, Panopto also enhanced the ability to brand its system without coding. Specifically, users can add three logos to the different-sized screens in the system like the player, library and the like by uploading different graphics to the system. While the three-logo schema may sound like overkill, you want a different-sized logo on these screens, and three separate logos is better than squishing one logo to fit. You can also choose an accent color for all screens.
Finally, Panopto shipped its Android app which scales to the screen size of the viewing device. Vieweres watching on Android tablets will get an experience close to that on computers, while smartphone viewers will get a single MP4 stream containing both slides and video.
The updated platform offers an end-to-end HLS video workflow. Panopto has also partnered on a rack-mounted video capture appliance.
This software-as-a-service solution lets companies and educators create libraries of on-demand content, or stream live to employees or students. We find it well-conceived, but difficult with longer files.
Companies or universities with massive video libraries will appreciate the ability to load hundreds of hours at once.
By combining lecture capture with mobile device interactivity, educators gain new ways to engage students in or out of the classroom.