As students around the world are transitioning to distance learning, an Israeli startup has inked a deal with Harvard University to make the classroom experience more productive.
Tel Aviv-based Verbit has developed software to transcribe audio into text automatically. Verbit says its artificially intelligent tech can transcribe audio 10 times faster than the competition for half the price – and at 99% accuracy.
Harvard will offer students both the original class video and the text transcriptions. This makes distance learning an attractive option compared with the classic frontal lecture where students must do their own rapid notetaking.
In January, Verbit announced that it had raised $31 million, bringing its total funding to $65 million. The company will use the latest funding to add additional languages to its service, which is now focused on English.
Harvard is not the only academic institution on Verbit’s growing customer list: Stanford, the University of Utah, the University of California Santa Barbara, Oakland University and the already-online Coursera all use Verbit. The company has 150 customers in total, with offices in Tel Aviv, New York and Kiev.
Verbit employs an acoustic algorithm to reduce background noise, identify speakers and cancel out other factors that reduce audio quality; a linguistic algorithm to identify domain-specific terms and recognize accents and dialects; and a contextual events algorithm that incorporates current events and the latest news to guarantee high precision.
The output is checked for accuracy by a human operator fluent in the relevant field of the conversation – be that law, insurance or mechanics – before files are uploaded to the Verbit cloud. Turn around can be as fast as four hours.
Verbit’s AI-powered software can also be used for creating video captions.
For more information, click here