Sogou Launches AI-Powered 'Simultaneous Interpretation 3.0'
CEO Insights team, Press Release
Sogou Simultaneous Interpretation 3.0 is the first of its kind that can ‘see’ and ‘think’ in real-time during the interpretation process. It applies leading Optical Character Recognition (OCR) and Natural Language Processing (NLP) technologies that enable the solution to see and process images on presentation slides, further interpreting the underlying information. To ‘think’ while interpreting, the solution utilizes Sogou's proprietary Context Engine, Knowledge Graph technology, and online encyclopedia Sogou Baike to generate contextual knowledge and collect extensive background information that in real-time comprehends the speaker's message accurately.
Sogou Simultaneous Interpretation 3.0 is the first of its kind that can ‘see’ and ‘think’ in real-time during the interpretation process
The solution analyzes sensory inputs from different modalities including vision, audio, and text, improving the recognition accuracy of slides-based content by 21.7 percent and lifting translation accuracy by 40.3 percent. The advancement, which marks another stride forward in the field of AI simultaneous interpretation, is powered by Sogou's multimodal cognition capacity. Moreover, the solution builds a comprehensive knowledge database by utilizing each presentation, thereby helps speakers with different expertise to explain specialized information more precisely.
Sogou Simultaneous Interpretation 3.0 recently made its debut at the GeekPark IF X conference in Beijing in December. Its AI interpretation technology has supported hundreds of domestic and international events, including the World Internet Conference, 2018 China Open Tennis Championship, and 2018 RISE Hong Kong Summit. With the latest breakthrough in multimodal cognition technology, Sogou looks to provide accurate and smooth interpretation services in more scenarios including conferences, video translation, and live broadcasts.