Simultaneous Interpretation for Face-to-Face Services

Toshiba has developed simultaneous machine interpretation technologies that support translation from Japanese to English, Chinese, and Korean, and vice versa, as well as a software interpretation system intended for people engaged in face-to-face services who need to communicate with people speaking those languages.

The simultaneous interpretation technologies consist of (1) an automatic speech recognition (ASR) technology that recognizes users' spontaneous and continuous utterances, and (2) a spoken-language machine translation (MT) technology that detects sentence boundaries and translates only the significant parts of sentences. These technologies are designed for cloud services and work together in a coordinated manner, enabling the system to interpret users' utterances within a short interval. Consequently, the system achieves an interpretation time 44% shorter than that of a conventional sentence-by-sentence interpretation system.

An evaluation experiment conducted in the city of Chiba showed an 81% interpretation success rate for predefined situations in tourist guidance and daily conversations.

TTS: text-to-speech

Outline of simultaneous machine interpretation for face-to-face communication