|
Hierarchical Labelling of Speech Databases
Steve Cassidy
Speech, Hearing and Language Research Centre, Macquarie
University
When: Tuesday, 24th July 1994
Time: 4:00pm
Where: Room E6A357, Macquarie University
Abstract:
The talk will describe the MU+ speech database system developed at SHLRC and used to access segments from a large corpus of spoken language. Each utterance in the database is described by a hierarchical labelling which defines the structure of the utterance and enables segments at any level --- eg. Phoneme, Syllable or Word -- to be extracted according to various conditions.
We have an existing software tool which generates this hierarchical labelling given a set of phonetic labels and the text of the utterance. However, we are now discovering that this tool is not sufficiently flexible, and cannot cope, for example, with utterances from other languages which have different syllabic structure.
I will describe how the hierarchical labels are currently generated and sketch out some ideas for a more flexible software system which is more adaptable to other languages and other structural models of language.
Enquiries: Maria Milosavljevic 9850 6345 mariam@mpce.mq.edu.au
| Last modified: July 1997 |