First International Conference on Spoken Language Processing (ICSLP 90)

Kobe, Japan
November 18-22, 1990

A Distributed Speech Database With an Automatic Acquisition System of Speech Information

Shozo Makino, Toshihiko Shirokaze, Ken'iti Kido

Research Center for Applied Information Sciences, Tohoku University, Sendai, Japan

This paper describes a distributed speech database with an automatic acquisition system of speech information. The first feature is a distributed speech database composed of various types of storages(optical disk, magnetic disk, CD-ROM) and communication networks(Ethernet, GP-IB). This database contains 12 GB speech waveform data. The second feature is a virtual database. An unified dictionary administrates allocation of the original speech waveform data and the secondary data on various kinds of storages. It is not necessary for users to know the allocation of the data. The third feature is the automatic acquisition system of speech information based on a given phoneme string for an input speech. The fourth feature is a user-friendly database editor. The database is built up based on relational tables. The database editor has functions of data insertion, data deletion, and data display. Furthermore, frequently used speech analysis methods, a speech synthesizer and a speech recognizer are implemented to the database. Using those functions it is very easy for an user to carry out speech research without any knowledge concerning to the database.

Full Paper

Bibliographic reference.  Makino, Shozo / Shirokaze, Toshihiko / Kido, Ken'iti (1990): "A distributed speech database with an automatic acquisition system of speech information", In ICSLP-1990, 1109-1112.