Naver Unveils New Voice Recognition Technology 'NEST'

[Asia Economy Reporter Buaeri] Naver unveiled its next-level speech recognition engine, ‘NEST’ (Neural End-to-end Speech Transcriber), on the 13th.

‘NEST’ is a technology that can accurately recognize and convert complex and diverse long-form speech expressions into text with limited data training. It enables precise speech recognition for unexpected expressions without pre-training on large amounts of refined data.

According to Naver, by improving the conventional modeling method that separately learns acoustic and linguistic information into an integrated modeling method (end-to-end), the amount and time of data required for training were reduced to about one-tenth of the previous level, while recognition accuracy was actually increased.

Naver’s new speech recognition technology is applied to the AI-powered ‘Clova Care Call’ service, which verifies active COVID-19 monitors via phone calls, and was also introduced in January to the automatic captioning service for Naver video news.

Naver plans to expand automatic captions using ‘NEST’ technology to various video and audio services and utilize the technology for video search and editors as well.

General users can experience the ‘NEST’ technology for free on the Clova Speech homepage, and companies and organizations can use it through partnership proposals. Starting with a Japanese release in the second half of the year, the supported languages will be expanded to include English, Chinese, and others.

Han Ik-sang, a Naver leader, stated, “‘NEST’ is expected to have high utility in various fields such as subtitle production and archiving of video and audio content, as well as call data management in customer centers,” adding, “We will continue to focus more on core AI technology research and further enhance the quality and efficiency of speech recognition.”

This content was produced with the assistance of AI translation services.