Fourth European Conference on Speech Communication and Technology

Madrid, Spain
September 18-21, 1995

The Dutch Polyphone Corpus

E. A. den Os (2), T. I. Boogaart (1), Lou Boves (1,2), Esther Klabbers (1)

(1) KPN Research, Leidschendam, The Netherlands
(2) SPEX, Leidschendam, The Netherlands

This paper first summarizes the work done to design, record, transcribe, and produce the Dutch Polyphone corpus. In addition, figures related to frequency of occurrence of diphones and triphones in the phonetically rich sentences are presented. Furthermore, we describe one way of using the corpus, viz. to derive information about the way callers pronounced telephone numbers, postal codes (both read and spontaneously spoken), amounts of money, times, and bank accounts. Finally, it is shown that hesitation phenomena occur most frequently within longer items.

Full Paper

Bibliographic reference.  Os, E. A. den / Boogaart, T. I. / Boves, Lou / Klabbers, Esther (1995): "The Dutch polyphone corpus", In EUROSPEECH-1995, 825-828.