Technology

Live translation: Zoom offers itself the start-up Kites

Zoom announced on Tuesday the acquisition of the company Kites, a German start-up behind a real-time machine translation platform. Kites’ team of 12 researchers will join the videoconferencing specialist’s workforce, with the ambition to improve meeting productivity through multilingual translation capabilities for Zoom users.

“With our aligned missions to make collaboration frictionless – regardless of language, location or other barriers – we are confident that the Kites team will fit seamlessly into Zoom,” said Velchamy Sankarlingam. , president of product and engineering at Zoom, at the end of this transaction, the amount of which was not disclosed.

Kites’ technology claims to be able to translate spoken language spontaneously, with minimum latency and maximum accuracy. The company claims that when it comes to a conversation, their system has an error rate of around 5%, while the error rate of human translation is around 5.5%.

Aborted trials

Enough to revolutionize an area of ​​passage from speech to text which has experienced its share of difficulties in recent years. In 2017, for example, Google released a highly anticipated new pair of wireless headphones, which boasted an exclusive real-time translation feature. Google’s pitch stated that its Pixel Buds could recognize speech in one language, translate the words into another language on a user’s phone, and then read the translated sentence aloud.

Alas, the first returns on this product were catastrophic. Technology had difficulty recognizing the words of speakers, especially if they spoke in complicated sentences or with an accent.

The problem lies in the fact that the recognition of human speech is difficult, regardless of the sophistication of the artificial intelligence, one explained on the side of the American giant. The American giant has put the cover back in 2020, by announcing a new feature to its Google Translate application, to allow the latter to transcribe audio from one language to another in near real time.

Source: .com

Back to top button