The web needs better video transcriptions. Around the world, we’re sharing more information than ever, and we’re sharing it with people with all kinds of abilities and disabilities. For the most part, video leaves a lot of people out.
Koemei is a big-thinking startup that’s working on programmatically transcribing video content.
Manual transcription, a.k.a. paying some poor schlub to schlock through every second of your content and painstakingly type out each word, is cost-prohibitive. Koemei’s solution is to first use an automated transcription program (à la Google Voice voicemail transcriptions, but far more accurate), then pull in the human intelligence to correct errors in a multi-user editing interface. Editing can be closed or public, depending on the material and the needs of the community using the video.
Best of all, the company has estimated it only takes one hour for its platform to turn around an entire hour’s worth of audio.
The transcriptions can be published to YouTube, Vimeo, and other video platforms, and (because it takes some hands-on experience before you’ll trust the impressive claims Koemei is making) the first 10 hours of transcription are free.
“The same thing Google has done for text, we’ll be able to do for videos, to let you search with text through your video libraries,” said Koemei founder Temitope Ola (pictured below) today as the company showed off its product at the DEMO conference in Santa Clara, Calif.
The startup estimates the market for video transcription at close to $16 billion annually and counts around 120,000 transcriptionists in the United States alone. Koemei also anticipates 21-percent year-over-year growth in the business, especially for corporate and educational video.
Koemei has offices in Martigny, Switzerland and San Francisco.
Koemei is one of 80 companies chosen by VentureBeat to launch at the DEMO Spring 2012 event taking place this week in Silicon Valley. After we make our selections, the chosen companies pay a fee to present. Our coverage of them remains objective.