A database of isolated spoken words has a 212‐word vocabulary uttered by 40 male and 30 female speakers and further a 3000‐word vocabulary uttered by 10 male and 20 female speakers. The 212‐word vocabulary was selected to contain most of the Japanese phoneme contexts. The 3000‐word vocabulary is a set of Japanese railroad station names. The speech data were sampled at 24 kHz and digitized to 12 bits. About 1000 magnetic tapes were used for the recording of the speech data. However, because of the huge number of the magnetic tapes, it is very difficult to deliver these tapes to speech researchers, and there are many problems in maintenance. For building up a database, the CD‐ROM has many merits such as compactness and price. Furthermore, the so‐called High Sierra Format was published last year and most computers can access CD‐ROMs. The magnetic tapes are being converted to CD‐ROMs in that format. By April 1989, two CD‐ROMs will be available that will contain the 212‐word vocabulary uttered by 50 speakers.