Third International Conference on Spoken Language Processing (ICSLP 94)
Most conventional applications for speech synthesis employ digitized speech messages. The voice quality is high but these applications have several weak points. For instance, the output voice messages are limited and when a new service item is added, it is expensive to reconstruct the speech file. The more cost-effective alternative is text-to-speech(TTS) technology but several problems are obvious in current systems. First, the author overviews the speech synthesis applications currently active in Japan; digitized speech applications and TTS applications are covered. The size and trends of the speech synthesis market are examined. Voice response applications in which digitized speech is basically used, are investigated. Furthermore, applications of TTS are estimated from Japanese TTS synthesizer shipments. For both speech synthesis techniques the author indicates the problems to be solved and proposes the research direction. In particular, TTS technology is focused on because it seems capable of greatly expanding the speech synthesis market. At present, TTS quality is still insufficient compared to digitized sound and the applications are restricted to a small domain. However, TTS technology possesses many advantages, all of which are well known, so if the speech quality can be improved, the number of application domains will rapidly increase.
Bibliographic reference. Hirokawa, Tomohisa (1994): "Trends in the applications of and market for speech synthesis technology", In ICSLP-1994, 675-678.