音声認識字幕における円滑な連携作業を実現するためのソフトウェア開発と情報保障者の技能(一般セッション,コミュニケーション支援,共生コミュニケーション及び一般)

概要

論文の詳細を見る
Automatic speech recognition (ASR) technology which has been the focus of attention in recent years is expected to be a realistic method that enables the captionists to follow the speed of delivery of the speaker's speech without being dependent on advanced skills. However, captioning using ASR technologies is not easy. Further, majority of captionists using existing methods are ordinary students without advanced skills. In general, captionists using ASR software, after listening to the speech of the speaker, "re-speak" (repeat what you heard exactly) the same content that was heard in a clear and distinct voice. The clear and distinct utterance that was repeated is sent to the ASR software installed in a personal computer that performs "speech-to-text conversion". In addition, two "corrector" are also needed to correct any errors and omissions. In order to achieve these two roles (Re-speaker and Corrector), a variety of equipment and software will also become necessary. We developed the supportive system that enables captionists who perform these "Re-speaking" and "Correcting" to perform their activities smoothly and with ease. Moreover, experienced knowhow based on the opinion that each role had reported was summarized. This paper reports the supportive system development and the experienced knowhow.
社団法人電子情報通信学会の論文
2009-05-07