Narodna in univerzitetna knjižnica, Ljubljana (NUK)
Naročanje gradiva za izposojo na dom
Naročanje gradiva za izposojo v čitalnice
Naročanje kopij člankov
Urnik dostave gradiva z oznako DS v signaturi
  • Prediction of symbolic prosody breaks with neural nets = Napovedovanje simboličnih prozodičnih mej z nevronskimi mrežami
    Stergar, Janez, 1968- ; Horvat, Bogomir, 1936-
    In this paper the data driven prediction of word level prosody modeling for Slovenian language is presented. Automatic learning techniques depend on the construction of a large corpus labeled with ... appropriate labels. The labeling can be done either automatically or by hand. While automatic labeling can be less accurate than hand labeling, the latter is very time consuming and in some cases inconsistent. Therefore we will present a new semiautomatic approach for determining prosody breaks features with a graphical user interface (GUI). The GUI combines the advantage of hand labeling and automatic labeling by achieving a high consistency in labeling and reducing the time that would be needed for hand labeling. The labeled Slovenian corpus has been used to train our phrase break prediction module, using a neural network (NN) structure. We used an MLP structure suitable for Digital Signal Processor (OSP) implementation. Experiments for the data driven prediction of major/minor phrase breaks have been performed. The achieved prediction accuracy marks a good entry-level for phrase break prediction of the Slovenian language and is comparable to other approaches in phrase break prediction where more complex prediction methods were used and a much larger corpus was used for training. The achieved overall prediction accuracy is about 90%.
    Vrsta gradiva - članek, sestavni del
    Leto - 2002
    Jezik - angleški
    COBISS.SI-ID - 7705366