To view this page ensure that Adobe Flash Player version 11.1.0 or greater is installed.

POST PRODUCTION Dialog-Based Search Accelerates Content Discovery for Anyone With a Media Library by Drew Lanham, Senior Vice President and General Manager, Nexidia Media and Entertainment Division I f you can’t find it, then you can’t reuse it. Those are a broadcaster’s words to live by as media libraries grow at exponential rates. But the need to find and reuse media assets is not confined to large, traditional broadcasters. Organizations of all types and sizes are finding that the ability to uncover media assets quickly and use them immediately is the key to creating new and compelling programming, fulfilling programming requirements, serving their constituents, and realizing revenue streams. The question is, “How?” The answer is, “phonetic-based dialog search.” First, Some Basic Background In order to fi nd media assets quickly, we need to make them searchable, and in order to do that, we typically “tag” them with metadata. This metadata can describe the content in several ways, but at a minimum it would include a fi lename with fi le type, timecode, and duration information. Traditional metadata might also include ratings, descriptions, or relevant keywords. When it comes to metadata, the more you have, the better your chances of fi nding exactly what you need. Metadata gets attached to the media in a process called logging, which is usually done manually. In the case of live programming, it can be done on the fl y so that the media is instantly searchable. For prerecorded content, logging can be a very labor-intensive, tedious, and potentially expensive process that can often lead to inaccurate metadata tags. The cost of manual logging can be $40-80 per hour, not to mention that it can be two to four times slower than real time. Transcription, at $1 per minute, is no better. Captioning can also be pricey at about $4 per minute. If an asset library has no metadata to begin with, such as one that has only recently been digitized, then inserting that metadata after the fact can be an even more daunting job — one that could overwhelm even the most well-funded and well-staffed media organization. It is for those reasons 56 | KITPLUS - THE TV-BAY MAGAZINE: ISSUE 97 JANUARY 2015 that many assets go without suffi cient metadata, and, in turn, go unused. Enter Dialog Searching It is against this backdrop that the benefi ts of implementing a phonetic-based dialog search tool become clear. Dialog searching not only makes a difference to major networks but to plenty of others, such as news and reality show producers, corporate or nonprofi t media departments, advertising departments, houses of worship, and sports distributors. One such tool is Nexidia Dialogue Search, a phonetic- based system that lets you search using plain language by phonetically indexing a media library. That process results in metadata that is searchable based not on the information that has been typed into the metadata fi elds during the logging process, but on what is actually spoken on the audio tracks. When you select an asset or library for indexing, software analyzes the audio tracks and creates a searchable index of the dialogue. Unlike speech- to-text methods, this system does not miss important keywords and proper names. Once the index is created, metadata is no longer required to search for an asset (though the system can indeed work in conjunction with any existing metadata). Suddenly it’s possible for anyone — from the largest broadcaster to the smallest nonprofi t media department — to search massive media libraries in seconds, and likely uncover assets they couldn’t fi nd for years or didn’t even know they had. Dialogue Search can work alone or in concert with the existing metadata in many media asset management systems, providing an additional layer of search power and increased accuracy. Use Cases That Go Beyond Broadcast Sports In sports distribution, the goal is to publish the content as quickly and as accurately as possible with as much value- added information as possible, such as instant replays, team and player statistics, highlight packages, and archive