In order to improve speech naturalness of a unit selection TTS system it is necessary to annotate prosodic phrase boundaries in the whole source corpus, which is extremely difficult to achieve manually. It is thus usefull to employ a machine classifier. This paper discusses suitable feature selection for such classification of a Czech TTS corpus, presents results of experiments with linear and quadratic classifiers and artificial neural networks, and compares them with human annotators.
Index Terms: speech synthesis, prosody, prosodic phrase, classification, neural network, unit selection, corpus
Cite as: Romportl, J. (2010) Automatic prosodic phrase annotation in a corpus for speech synthesis. Proc. Speech Prosody 2010, paper 892
@inproceedings{romportl10_speechprosody, author={Jan Romportl}, title={{Automatic prosodic phrase annotation in a corpus for speech synthesis}}, year=2010, booktitle={Proc. Speech Prosody 2010}, pages={paper 892} }