An object-oriented speech database system is described that allows automatic persistent storage and updating of almost arbitrary data types and relations describing speech. The system is programmed in Common Lisp and CLOS for flexibility and easy extendibility. It allows for powerful search, analysis, and modification of database contents. A phonetic representation framework for Finnish is available where the hierarchical structure of speech is easily manipulated. Deferred loading and caching of data to working memory allows for fast computation and effective memory usage even with very large databases. A graphical user interface supports manual interaction and transcription tools. A semiautomatic transcription system is under development.
Bibliographic reference. Karjalainen, Matti / Altosaar, Toomas (1993): "An object-oriented database for speech processing", In EUROSPEECH'93, 183-186.