Corporate Research & Development Center

Toshiba has developed a content transliteration system called "DaisyRings™", which allows users to create spoken content easily using text-to-speech technology. Users can correct the text and accent information easily on a web browser. The system is designed for librarians, volunteers and teachers to help them make and edit content for the visually impaired.

With e-books gaining popularity, book-reading utilizing text-to-speech technology is likely to be used for assisted reading for the visually impaired. However, e-book contents created for reading and display have various problems of misreading, wrong accents and reading without pauses. Moreover, transliteration of a book by read it aloud by volunteers is laborious and time-consuming.

Toshiba has developed DaisyRings™, a text-to-speech technology-based content transliteration support system to assist the transliteration of large quantities of books and specialized books, which is difficult to do by human narration. The user can upload a text for transliteration and download the edited content as a voice file and DAISY-formatted file. The system is easy to use and the text and accent information can be corrected by inputting "ruby-type" data. To make a pause, the user only needs to insert a code.
The user can choose from four types of high-quality artificial voices (two men and two women) developed by Toshiba to make the narration sound natural and the reading experience more enjoyable.
The conventional method of transliterating and editing a 150-page text takes about one and a half months, but the new system can shorten the process to only one week. The system has been used for free by volunteers and librarians for one year since November 2013 (Note), and by the Japan Braille Library since December 2013.

We plan to apply the technology to voice content creation businesses and continue to improve the text-to-speech technology to satisfy various needs.