Human Science

Research on automatic captioning of news programs

Advanced Imaging Devices

Research on ultrahigh sensitivity/ultrahigh definition imaging devices
Demands to expand closed-captioning of news programs are received from the elderly and the hearing impaired. Our group has been promoting research to provide real-time automatic closed-captioning of news programs utilizing speech recognition technology.
News programs cover topics from a wide variety of fields such as politics, economics, and sports, in addition to using new biographical and geographical names and words from foreign languages. These need to be accurately recognized and automatically generated by a closed-captioning system in proper Japanese, using kanji and kana. This requires development of an advanced speech recognition system.
As for the speech recognition performance, an accuracy rate exceeding 95% within a response time of less than two seconds is the research goal that we are striving toward.
We have been conducting comprehensive research and development for imaging materials and devices for next generation broadcasting systems, with application to HDTV and its succeeding technology.
The well-known camera with a HARP photoconversion film has been utilized to the full due to its ultrahigh sensitivity advantage, and has played a conspicuous part in programming, especially for emergency news and science programs. Research and development is now underway on two image devices which inherit the superior performance of HARP technology. One is a solid-state HARP image sensor which aims to improve the reliability and mobility of HDTV cameras. The other is a field emitter image sensor with HARP film, which is hoped to bring ultrahigh definition/ultrahigh sensitivity to next generation broadcasting systems. Besides this, basic research on new photoconversion materials that are to follow HARP technology is also being promoted.
Haruo ISONO,
Human Science Division
Masahide ABE,
Advanced Imaging Devices Division