How speed search can bring the full value of video to enterprises

Speed search
Fast talking

Video usage as a communications tool is on the rise and firms are amassing more and more knowledge and valuable content on video. The concept of a 'Corporate YouTube' has become a reality and this is about to change the way we search for and consume content while at work.

Yet, how easy is it to find this valuable content once it has been created, already viewed and archived? Not just the particular video you are looking for but the specific, spoken content within a video. Tagging and metadata is helpful yet time consuming and it is often not specific enough for an accurate search.

Imagine being able to find your favourite videos with simple key phrases or comments you heard, no matter what device you are on, be it mobile or not. Whether it is a conference, webinar, expert video or just any general 'How To' video you really liked, a speech search functionality can help you find that specific video instantly. Now let me take that context within the workplace.

A video platform which offers speech search functionality provides the technology to index the audio of all videos within a large media library and allows for any word or sentence to be searched – across thousands of hours. A phonetic search will deliver accurate results across different speakers, accents and across recordings of varying degrees of quality. This reduces the time it takes for users to not only find the particular video they are looking for but it also makes it possible for them to search for key words so that they can ultimately locate the precise section of the video containing the data they need.

Speech search can dramatically improve the way you look for information within a video and removes completely the task of combing through whole, long videos leaving you more time to focus on the business and on the task of learning and digesting the content.

A wholesome speech search technology should feature phonemes analysis instead of words for increased accuracy over speech to text methods. It should also allow for proper names and slang recognition to prevent individual accents to cause any confusion.

Key elements should be the capability to quickly preview results without the need to scroll through hour-long clips just to find a 1- minute topic but also the presence of confidence scores to allow you to focus only on the closest matches or to broaden your search for the inclusion of subtle variations.

Language packs should be installed to allow for multiple language search and there should also be the option to search not only single videos but series of videos so that dialogues can be found on top of spoken words.

Last but not least, speech search should be linearly scalable to support even the largest video libraries with full redundancy.

A speech search functionality can benefit organisations by making it possible for knowledge to be found – and therefore absorbed – fast but also by helping to keep content secure, managed and monitored.
In conclusion, what might in the past have taken hours of video library searching, watching and listening to find, now takes only minutes. If it is spoken, you can find it via speech search quickly and accurately.

  • Martin Nurser joined Qumu in 2013 and brings more than 25 years of enterprise and technology industry experience with emphasis on engineered systems and cloud technologies. He is a leading expert in channel and strategic alliance development and management, go-to-market planning, as well as market segmentation and business strategy formulation.