For many people the vacation season will look totally different this 12 months, separated from the individuals we love. Should you’re on this boat too—mitigating the unfold of the coronavirus—thanks and we hope the next story would possibly supply another, however useful method to join with family and friends.
Whereas we all know digital get-togethers can by no means absolutely match the intimacy of in-person conversations, they’ll maintain us related and possibly even protect some particular moments for future generations.
On this spirit, we’re sharing our collaboration with StoryCorps, a nationwide non-profit group devoted to preserving humanity’s tales by means of 1:1 interviews. Over the previous 17 years, StoryCorps has recorded with greater than 600,000 individuals and despatched these recordings to the U.S. Library of Congress the place they’re preserved for generations to come back on the American Folklife Middle.
That is the world’s largest assortment of human voices on the planet, however, it’s been comparatively inaccessible. That’s when StoryCorps approached us to assist make its wealthy archive of first-person historical past universally accessible and helpful.
StoryCorps + Google Cloud AI
In 2019, StoryCorps and Google Cloud partnered to unlock this superb archive utilizing synthetic intelligence (AI) and create an open, searchable and accessible audio database for everybody to search out and hearken to first-hand views from humanity’s most necessary moments.
Diving into how this works: for an audio recording to be searchable, the audio file and “moments” or key phrases inside that file—wanted to be tagged with phrases for which you’d search.
First we used Speech-to-Text API to transcribe the audio file.
Then Natural Language API recognized key phrases and their salience from the transcription.
The transcript and key phrases have been loaded to an Elastic Search index.
Leading to a searchable transcript on the StoryCorps Archive.