A Welsh speech database: preliminary results.

Abstract

A speech database for Welsh was recorded in a studio from read text by a few speakers. The purpose is to investigate the acoustic characteristics of Welsh speech sounds and prosody. It can also serve as a resource for future work in speech synthesis and recognition. The speech is labelled by hand at the acoustic phonetic level, and labelled semi-automatically at the phoneme, syllable, and word levels. Statistical analysis of the database resolves some long-standing questions in the phonetics of Welsh stress, and yields more data on Welsh speech sounds. The overall procedure could be useful for work on other minority languages where very little basic acoustic phonetic research has yet been done.