Launching technology to convert voice... into text

The Electronics and Telecommunications Research Institute (ETRI) of Korea claims to have developed a translation technology that uses audio recognition to "translate" content from one language to another.

ETRI researchers have applied deep learning technology of artificial intelligence to develop a multilingual audio recognition-translation program.

ETRI staff test their speech recognition technology using the Genie Talk mobile translation app on April 18. (Photo: ETRI)

Accordingly, with available phrases, the technology will recognize voice sounds to translate directly from one language to another and then display that content in writing.

Compared to Google's program, ETRI's Korean recognition ability is superior, being able to "translate" up to 96% of the content of daily conversations. This rate in other languages ​​is from 90% to 95%, equivalent to Google.

The technology is currently available in nine languages, including English, Chinese, Japanese, Spanish, French, German, Russian and Arabic.

ETRI plans to expand its application to several languages ​​in Southeast Asia.

According to Khoahoc.tv